Table 1.

Recommendations for management by the intensivist of gastrointestinal bleeding

Area 1: Management of GI bleeding before endoscopic diagnosis		11	In children, in massive hematochezia and/or with hemodynamic consequences, and when GI endoscopic findings are normal, an emergency scintigraphy to search for a Meckel’s diverticulum should be used and/or a surgical exploration (McBurney’s incision or coelioscopy) (strong agreement)	5	In the presence of stigmata associated with a high risk of rebleeding (Forrest type Ia, Ib, IIa, IIb), PPI treatment should be continued at “high” doses for 72 h (strong agreement)
1	Nasogastric intubation may help confirm, but cannot discount, suspected upper GI bleeding (strong agreement)	12	Vasoactive treatment (terlipressin or somatostatin derivative) should be administered as soon as possible when portal hypertension is the suspected cause if GI bleeding (strong agreement)	6	Second-look EGD should not be done routinely (strong agreement)
2	Suspected rupture of esophageal/gastric varices probably does not contraindicate nasogastric intubation (strong agreement)	13	In a patient already treated with noradrenaline, specific vasoactive treatment of the splanchnic area (terlipressin, somatostatin, somatostatin derivative) should probably be administered when portal hypertension is the suspected cause of GI bleeding (expert opinion, weak agreement).	7	Second-look EGD should probably be done when a high-risk stigmata has been observed(weak agreement)
3	To ensure emptying of the stomach content before EGD, intravenous erythromycin should be administered at a dose of 250 mg (5 mg/kg in children), in the absence of contraindications (strong agreement)	14	Specific vasoactive treatment of the splanchnic area (terlipressin, somatostatin, somatostatin derivative) should probably not be administered when portal hypertension is not the suspected cause of GI bleeding (weak agreement)	8	Patients with ulcer bleeding should not be treated with H2 receptor antagonists (strong agreement)
4	If a nasogastric tube has been inserted, gastric lavage to empty the stomach is an alternative to administration of erythromycin (weak agreement)	15	In GI bleeding potentially caused by ulcers, PPI treatment should be started without waiting for endoscopic diagnosis (weak agreement)	9	In adults, in case of Forrest type Ia and Ib, first- intention selective arterial embolization by interventional radiology should probably be used following failure of endoscopic therapy (weak agreement)
5	In adults, the Rockall score and the Glasgow-Blatchford bleeding score can probably help to identify patients at high risk of morbidity and mortality and to refer them to an intensive care unit (strong agreement)	16	In GI bleeding potentially caused by ulcers, high-dose PPI treatment should probably be administered (weak agreement)	10	In adults, in case of Forrest type Ia and Ib and catastrophic bleeding, first-intention surgical hemostasis should probably be used following failure of endoscopic therapy if local conditions do not allow arterial embolization(strong agreement)
6	EGD should be done in the 24 h following the admission of the patient with suspected upper GI bleeding (strong agreement)	Area 2: Treatment of upper GI bleeding unrelated to portal hypertension		11	Biopsy screening for Helicobacter pylori infection can be performed during the first EGD for GI bleeding without worsening the bleeding (strong agreement)
7	EGD should be probably done in the b12 h following the admission of the patient with suspected esophageal/gastric variceal bleeding (strong agreement)	1	In the presence of stigmata associated with a low risk of rebleeding (Forrest type IIc and III), endoscopic hemostasis should not be used (strong agreement)	12	There is probably no advantage to emergency treatment of Helicobacter pylori infection in the case of GI ulcer bleeding (strong agreement)
8	EGD should be probably done as soon as possible, and once the patient is resuscitated, when active upper GI bleeding is suspected (strong agreement)	2	In the presence of stigmata associated with a low risk of rebleeding (Forrest type IIc and III), PPI treatment at “standard” doses should be continued (strong agreement)	13	Aspirin antiplatelet therapy should probably be maintained in the case of GI ulcer bleeding until consultation with specialists (weak agreement)
9	In massive hematochezia and/or hemodynamic consequences, an EGD should be performed as soon as possible (strong agreement)	3	In the presence of stigmata associated with a high risk of rebleeding (Forrest type Ia, Ib, IIa), endoscopic hemostasis should be performed (strong agreement)	14	In dual antiplatelet therapy, clopidogrel should probably be stopped in the case of ulcer bleeding until consultation with specialists (strong agreement)
10	In adults, in massive hematochezia and/or with hemodynamic consequences, a CT angiography should be performed in emergency, if EGD is not rapidly available and/or if an aortoenteric fistula is suspected (strong agreement)	4	In the presence of an adherent clot (Forrest type IIb), endoscopic hemostasis is possible when the clot is small (strong agreement)	Area 3: Treatment of upper GI bleeding related to portal hypertension
1	Endoscopic therapy of bleeding esophageal/gastric varices should be done during initial EGD (strong agreement)	13	Blood transfusion in most patients should probably target a hemoglobin concentration of 7 to 8 g/dL (strong agreement)	5	When lower GI bleeding is massive, surgical hemostasis should be proposed in case of arterial embolization or colonoscopy failure or rebleeding (strong agreement)
2	Endoscopic therapy of bleeding esophageal varices is based on band ligation. Sclerotherapy is an alternative in the very young child (strong agreement)	14	In GI bleeding in patients with cirrhosis, there is probably no indication for administration of fresh-frozen plasma with the objective to correct a coagulopathy (strong agreement)	6	When lower GI bleeding is catastrophic, surgical hemostasis should be performed if arterial embolization is not possible under local conditions (strong agreement)
3	Endoscopic therapy of bleeding gastric varices is based on obturation with tissue adhesives (strong agreement)	15	In GI bleeding in patients with cirrhosis, there is no indication for administration of fresh-frozen plasma before EGD (strong agreement)	7	In massive or persistent lower GI bleeding, the small intestine should be examined as soon as possible when CT angiography and colonoscopy fail to locate the source of bleeding (strong agreement)
4	Vasoactive treatment should be continued for 3 to 5 days after endoscopic therapy of esophageal/gastric varices rupture (strong agreement)	16	In GI bleeding in patients with cirrhosis, there is no indication for the administration of factor VIIa (strong agreement)	Area 5: Prevention of upper GI bleeding in intensive care
5	In adults, after endoscopic hemostasis of bleeding related to portal hypertension, the placement of a TIPS within 72 h should be considered in high-risk patients (strong agreement)	17	In GI bleeding in patients with cirrhosis, platelet transfusion should probably be considered when bleeding is uncontrolled and platelet count is <30,000/mm3(strong agreement)	1	Patients with a history of peptic ulcer admitted to intensive care should probably be considered at risk of GI bleeding (strong agreement)
6	Balloon tamponade should be considered after endoscopy failure pending radical treatment of portal hypertension. In child, its use should probably be envisaged if emergency EGD is not possible (strong agreement)	18	In adults, in esophageal/gastric bleeding, beta-blocker treatment should be started when vasoactive treatment is discontinued (strong agreement)	2	Early enteral feeding is effective in preventing “stress ulcer” bleeding (strong agreement)
7	Antibiotic prophylaxis with third-generation cephalosporin or with fluoroquinolone for 5 to 7 days should be given to any cirrhotic patient with GI bleeding (strong agreement)	19	After ligation of esophageal varices, nasogastric intubation should probably be avoided (expert opinion, strong agreement)	3	Patients requiring mechanical ventilation for more than 48 h and for whom enteral feeding is not possible should be considered to be at risk of “stress ulcer” bleeding (strong agreement)
8	Lactulose treatment to prevent hepatic encephalopathy should probably not be initiated during GI bleeding in a cirrhotic patient (expert opinion, strong agreement)	Area 4: Management of presumed lower GI bleeding		4	Patients admitted to intensive care with kidney failure and/or coagulopathy and/or receiving antiplatelet therapy should be considered to be at risk of “stress ulcer” bleeding (strong agreement)
9	In adults, PPI therapy should not be initiated or continued when EGD has confirmed a diagnosis of ruptured esophageal/gastric varices (strong agreement)	1	In adults with massive hematochezia, demonstration of active bleeding by abdominal CT angiography or arteriography justifies embolization as first-line therapy (strong agreement)	5	Routine drug prophylaxis of “stress ulcer” should not be used in intensive care patients with enteral feeding (strong agreement)
10	In children, PPI therapy should probably be initiated or continued in case of esophageal/gastric varices rupture (weak agreement)	2	In massive hematochezia, and in the absence of detectable bleeding on CT angiography or arteriography a prepared colonoscopy should be performed within 24 h (strong agreement)	6	Ulcer prophylaxis medication should probably be given routinely in intensive care patients with a history of peptic ulcer (even if enterally fed) (weak agreement)
11	One objective of hemodynamic treatment during esophageal/gastric varices rupture should be to restore a satisfactory mean blood pressure to preserve tissue perfusion (strong agreement)	3	In massive and persistent hematochezia, and in the absence of detectable bleeding on abdominal CT angiography or arteriography, a prepared colonoscopy should probably be done within 12 h with the objective of performing endoscopic hemostasis (strong agreement)	7	Ulcer prophylaxis medication should probably be given routinely to intensive care patients receiving antiplatelet therapy (even if enterally fed) (weak agreement)
12	In adults, during esophageal/gastric varices rupture, early hemodynamic treatment should probably maintain mean blood pressure at approximately 65 mmHg in most patients (strong agreement)	4	In adults with massive hematochezia, rectosigmoidoscopy should probably be done if full colonoscopy cannot be performed within 24 h (strong agreement)	8	In the absence of enteral feeding, ulcer prophylaxis medication should probably be given to ventilated patients (weak agreement)
9	In the absence of enteral feeding, ulcer prophylaxis medication should probably be given to patients with coagulopathy (weak agreement)	13	A large bore nasogastric tube for aspiration should probably be replaced by a small-calibre enteral tube as soon as possible (expert opinion, strong agreement)
10	In children, a pediatric risk of mortality score (PRISM) > 10 associated with respiratory failure or coagulopathy or both probably calls for ulcer prophylaxis (strong agreement)	14	Antacids should not be used to prevent “stress ulcer” bleeding (strong agreement)
11	Screening for Helicobacter pylori should not be routine in intensive care patients (strong agreement)	15	H2 receptor antagonists and PPIs are probably comparable but of low efficacy in preventing “stress ulcer” bleeding (weak agreement)
12	A nasogastric tube should probably be removed once it is no longer used (expert opinion, strong agreement)	16	H2 receptor antagonists and PPIs are probably comparable regarding the risk ventilator associated pneumonia during mechanical ventilation (strong agreement)