Steven Selub, Jamie S. Barkin

Research output: Contribution to journalArticlepeer-review


Conn et al. (1) presented the results of a multicenter, randomized, double‐blind clinical trial to assess the efficacy of propranolol in preventing the initial hemorrhage from esophagogastric varices. One hundred and two patients (73 men, 29 women, mean age 54 yr) with the clinical diagnosis of cirrhosis (80 with alcoholic cirrhosis) were randomized to receive either placebo or propranolol. The patients had endoscopically documented esophageal varices and portal hypertension, and had not bled previously from varices or from an unknown upper gastrointestinal site. Patients with known neoplasms or severe hepatic disease or other medical conditions that might interfere with the study were excluded. Patients who agreed to undergo hepatic vein catheterization and whose hepatic vein pressure gradient (HVPG) was greater than or equal to 12 mm Hg were titrated with propranolol to reduce their HMPG to less than 12 mm Hg. After the dosage of propranolol had been thus determined, the patients were randomized to receive either propranolol (51 patients) or placebo. The double‐blindedness of the investigation was maintained by having the patients examined at visits by nurses or fellows. The dosages were not increased during the study. Patients returned monthly for 3 months, and every 3 months thereafter. Pill counts were made, and a compliance index was calculated, on the basis of the number of appointments kept, percentage of tablets consumed, propranolol levels, alcohol consumption diary, and blood alcohol levels. Heart rate was not used to measure compliance. The patients underwent endoscopy and hepatic vein catheterization at 3 and 12 months after randomization, and annually thereafter. Varices were measured endoscopically and graded according to size: grade 1, 1–3 mm with Valsalva; grade 2, 1–3 mm without Valsalva; grade 3, 3–6 mm; and grade 4, greater than 6 mm. Patient's Child's class was calculated by the Childs‐Turcotte criteria, with class A defined as a score of 5 to 8, class B as a score of 9 to 11 and class C as a score of 12 to 15. The investigators used the following endpoints: 1. Upper gastrointestinal hemorrhage, which was defined as hematemesis or melena that reduced the hematocrit by 6% or required blood transfusions. Variceal hemorrhage, documented by endoscopic visualization of active bleeding or fresh clot or eschar on the surface of a varix in the absence of any other possible upper gastrointestinal bleeding site. The examination was performed during active hemorrhage or within 24 h after bleeding stopped. The endoscopists were blinded to the patient's therapy. 2. Survival, which was expressed in number of months after randomization. The cause of death was determined in conference by the principle investigators who were unaware of the patient's therapy. 3. Complications, which ware defined as adverse events that may have been caused by propranolol and that necessitated stopping treatment. Decisions to stop therapy and recording of complications were done without knowledge of therapy. During the study period, from October 1982 to September 1986, approximately 40% of the cirrhotic patients admitted to the participating hospitals met the inclusion criteria. Of those excluded, 10% had bled before they could be assessed and randomized, 20% had contraindications to beta‐blockade or were already receiving it, 15% refused either endoscopy or catheterization or declined to participate, 10% were deemed noncompliant hy the investigators, and 5% had an HMPG less than 12 mm Hg. AH patients were followed until the last randomized patient had been treated for at least 6 months; mean follow‐up was 16.3 ± 12 months for the placebo group and 17.1 ± 10.9 months for the propranolol group. Eleven (22%) patients of the 51 patients in the placebo group bled from varices, compared with two (4%) of the 51 patients in the propranolol group, a difference that was significant (p < 0.01, relative risk 1.4 to 32.2, 95% CI). The difference was more pronounced in patients with large varices; nine of 22 patients in the placebo group bled, whereas none of 27 patients with large varices in the propranolol group hied (p < 0.001). Patients with small varices exhibited no difference between placebo and propranolol treatment; eight percent bled in each group. Three patients in the placebo‐treated group bled from portal hypertensive gastropathy, whereas no patients in the propranolol group bled from this lesion. The effect of propranolol in preventing hemorrhage was confined nearly exclusively to alcoholic eirrhotics. Ten of 41 patients with alcoholic cirrhosis (24%) in the placebo group bled, compared witb two of 39 patients (5%) in the propranolol group (p < 0.01). Of the nonalcoholic eirrhotics, one of 10 in the placebo group bled, whereas none of the 12 in the propranolol group bled (NS). There was no significant difference in the bleeding rate in Child's group A or C patients, although the latter group was small. Among Child's class B patients, there was a significantly higher rate of hemorrhage in those receiving placebo, compared with propranolol treatment (nine of 24 vs. one of 11, p < 0.05). Among patients who took more than 75% of their tablets, eight of 35 (23%) in the placebo group bled, whereas only two of 39 (5%) in the propranolol group hemorrhaged (p < 0.05). There was no significant difference between groups, among less compliant patients. Survival analysis was based on intention to treat and included all patients. There were 11 deaths (22%) in the placebo group and eight (16%) in the propranolol group (NS). Cumulative survival was slightly but not significantly improved (Kaplan‐Meier) in the propranolol group. There were three deaths from varieeal bleeding in the placebo group and two in the propranolol group. Three patients in the placebo group (6%) and seven in the propranolol group (14%) suffered adverse effects requiring cessation of therapy (NS). Two patients in the propranolol group developed congestive heart failure, one developed bronchospasm, and one developed diabetes mellitus.

Original languageEnglish (US)
Pages (from-to)253-255
Number of pages3
JournalThe American journal of gastroenterology
Issue number2
StatePublished - Feb 1992

ASJC Scopus subject areas

  • Hepatology
  • Gastroenterology


Dive into the research topics of 'BETA‐BLOCKERS IN PREVENTION OF VARICEAL HEMORRHAGE: TANTALIZING EXPECTATIONS'. Together they form a unique fingerprint.

Cite this