Fort 1990.

Methods	Randomised: yes, by a table of random numbers. Concealment: method not described. Double‐blind: no. Description of withdrawals/dropouts: no Length of FU: until discharge. Jadad score: 2/5.
Participants	Country: France. Diagnosis of bleeding: by endoscopy. Source of bleeding: 100% confirmed or unconfirmed varices. Endoscopy performed in all patients prior to randomisation. Diagnosis of cirrhosis: Clinical or histological (PC). 34 patients with 47 episodes. Child's A = 15%, B = 30%, C = 55%. Alcoholic cirrhosis = 68%.
Interventions	Terlipressin versus balloon tamponade. R x 1: terlipressin 2 mg iv once, then 1mg iv every six hours for 30 h plus NTG 0.4mg. SL every 20 min for the first 6 h. R x 2: Sengstaken‐Blakemore tube for 24 h. Time from admission to treatment: unknown. Initial sclerotherapy: none.
Outcomes	1. Mortality at discharge. 2. Failure of initial haemostasis at 12 h. 3. Rebleeding 12‐48 h. 4. Procedures required for haemostasis: PC 5. Blood transfusions. 6. LOS: NR.
Notes	Source of funding: Included in hospital costs (PC). Patients in either group who continued to bleed at the end of 12 h received the alternative treatment (i.e., patients in the terlipressin group received balloon tamponade and vice versa).
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	B ‐ Unclear