Restrictive vs liberal transfusion for upper gastrointestinal bleeding: A meta-analysis of randomized controlled trials

Table 1 Characteristics of included studies

Ref.	Design	Regions	Target population	Intervention	Outcome measures
Blair et al[25]	Single center RCT	London, United Kingdom	Acute severe UGIB; no esophageal varices	Early transfusion group (liberal transfusion): at least 2 units of blood No transfusion group (restrictive transfusion): no blood transfusion unless hemoglobin fell below 8 g/dL or they were shocked	Observed results: coagulation results (impedance clotting time, kaolin cephalin clotting time); haematological results (hematocrit); eventual blood transfused; number of rebleed; number of death
Elizalde et al[24]	Single center RCT	Barcelona, Spain	Liver cirrhosis with an acute variceal bleeding episode	PRC group (liberal transfusion): 2 units of packed red cells PPL group (restrictive transfusion): 500 mL of a 5% plasma protein solution	Observed items: hemodynamic measurements (cardiopulmonary pressures, cardiac output, wedged and free hepatic venous pressures, mean arterial blood pressure, systemic vascular resistance); hormonal measurements (plasma renin activity, aldosterone levels, norepinephrine); rheological parameters (plasma volume, blood viscosity)
Villarejo et al[23]	Single center RCT	Buenos Aires, Argentina	Acute digestive hemorrhage with stable haemodynamics	Treatment group (restrictive transfusion): patients underwent normovolemic hemodilution with crystalloid solutions, and the hematocrit value was maintained as ≥ 21% and < 28%; red cell transfusion was given if angina, shock, hemodynamic instability, or hematocrit < 21% Control group (liberal transfusion): the target of transfusion in these patients was the hematocrit value of ≥ 28%	Observed results: organ failure, hospital stay, APACHE II score, red cell transfused, hematocrit, haemoglobin
Villanueva et al[14]	Single center RCT	Barcelona, Spain	Upper gastrointestinal bleeding	Liberal transfusion: the hemoglobin threshold for transfusion was 9 g/dL, with a target range for the post-transfusion hemoglobin level of 9-11 g/dL Restrictive transfusion: the hemoglobin threshold for transfusion was 7 g/dL, with a target range for the post-transfusion hemoglobin level of 7-9 g/dL	Primary endpoints: the rate of death from any cause within the first 45 d Secondary endpoints: the rate of further bleeding and the rate of in-hospital complications

PPL: Plasma protein solution; PRC: Packed red cells; RCT: Randomized controlled trial; UGIB: Upper gastrointestinal bleeding.

Table 2 Eligibility criteria of included studies

Ref.	Eligibility criteria
Blair et al[25]	Consecutive patients with acute severe upper gastrointestinal haemorrhage were prospectively randomized on arrival to receive during their first 24 h in hospital
	Only known cases of oesophageal varices were excluded as they frequently have abnormal coagulation due to liver disease
Elizalde et al[24]	The study population consisted of patients with cirrhosis of the liver admitted for the management of an acute variceal bleeding episode
	Only patients in whom hemostasia had been achieved within the previous 24-72 h by means of endoscopic sclerotherapy, and who were still anemic (hematocrit < 30%) and normovolemic as defined by clinical parameters (systolic pressure > 100 mmHg, right atrial pressure > 2 cm H2O, heart rate < 100 beats/min, and urine output > 0.5 mL/kg per hour) were eligible for the study
	Age < 18 or > 80 yr, renal failure as defined as serum creatinine > 2 mg/dL, portal thrombosis, diffuse or multinodular hepatocellular carcinoma, arterial hypertension, peripheral vasculopathy, previous surgical or transjugular intrahepatic portosystemic shunt, bacterial infection, and use of vasoactive drugs to prevent or treat portal hypertension-related bleeding were considered exclusion criteria
Villarejo et al[23]	Inclusion criteria: acute high digestive haemorrhage with stable haemodynamics and any aetiology; age > 15 yr old
	Exclusion criteria: history of angina; shock not responsive to volume expansion; requirement of surgery; established renal insufficiency; poliglobulina; bleeding diathesis; acute or chronic liver dysfunction; chronic anemia; pregnancy; sepsis; acute or chronic respiratory failure; haematocrit < 20% on admission; religious objection to transfusion
Villanueva et al[14]	Patients older than 18 yr of age who had hematemesis (or bloody nasogastric aspirate), melena, or both, as confirmed by the hospital staff, were considered for inclusion
	Patients were excluded if they declined to undergo a blood transfusion
	Additional exclusion criteria: massive exsanguinating bleeding, an acute coronary syndrome, symptomatic peripheral vasculopathy, stroke, transient ischemic attack, or transfusion within the previous 90 d; a recent history of trauma or surgery; lower gastrointestinal bleeding; a previous decision on the part of the attending physician that the patient should avoid specific medical therapy; and a clinical Rockall score of 0 with a hemoglobin level >12 g/dL

Table 3 Baseline characteristics of patients in included studies

Ref.	Groups	No. Patients	Age (yr)	Sex (male:female)	Source of bleeding	Hematocrit atadmission	Hemoglobin atadmission (g/dL)
Blair et al[25]	Restrictive transfusion	26	60 ± 4.5	2:1	Gastric ulcer (n = 4); duodenal ulcer (n = 13); carcinoma (n = 2); Mallory-Weiss tear (n = 3); not visualized (n = 4)	29 ± 1.6	NA
	Liberal transfusion	24	64 ± 3.6	2:1	Gastric ulcer (n = 2); duodenal ulcer (n = 17); carcinoma (n = 1); Mallory-Weiss tear (n = 2); not visualized (n = 2)	28 ± 1.2	NA
Elizalde et al[24]	Restrictive transfusion	8	60 ± 4	5/3	Bleeding from esophageal varices (n = 7); bleeding from gastric varices in the fundus of the stomach (n = 1)	27.0 ± 1.3	91.5 ± 6.8
	Liberal transfusion	8	64 ± 2	4/4	Bleeding from esophageal varices (n = 7); bleeding from gastric varices in the fundus of the stomach (n = 1)	27.0 ± 1.3	90.5 ± 3.96
Villarejo et al[23]	Restrictive transfusion	14	56.8 ± 12.8	9/5	Mallory Weiss (n = 2); erosive gastritis (n = 14); erosive gastroduodenitis (n = 4); Forrest gastric ulcer (n = 10); Forrest duodenal ulcer (n = 10); erosive duodenitis (n = 1)	26.9 ± 4.29	8.76 ± 2.47
	Liberal transfusion	13	45.3 ± 14.6	9/4		28.3 ± 5.59	8.85 ± 2.53
Villanueva et al[14]	Restrictive transfusion	444	NA	NA	Peptic ulcer (n = 228); gastroesophageal varices (n = 101); Mallory-Weiss tears (n = 25); erosive gastritis or esophagitis (n = 38); neoplasms (n = 16); other (n = 36)	NA	9.6 ± 2.2
	Liberal transfusion	445	NA	NA	Peptic ulcer (n = 209); gastroesophageal varices (n = 109); Mallory-Weiss tears (n = 30); erosive gastritis or esophagitis (n = 29); neoplasms (n = 20); other (n = 48)	NA	9.4 ± 2.4

Table 4 Quality assessment of included studies

Ref.	Randomization	Blinding	Withdrawals and dropouts	Jadad Score
Blair et al[25]	Yes (inadequate)	No	Unclear	1
Elizalde et al[24]	Yes (inadequate)	Unclear	Unclear	1
Villarejo et al[23]	Yes (adequate)	Unclear	Clear	3
Villanueva et al[14]	Yes (adequate)	Unclear	Clear	3