Current issues of ACP Journal Club are published in Annals of Internal Medicine


Review: β-Blockers reduce first-time variceal hemorrhage but do not improve 2-year mortality

ACP J Club. 1991 Sept-Oct;115:49. doi:10.7326/ACPJC-1991-115-2-049

Source Citation

Poynard T, Calès P, Pasta L, et al. Beta-adrenergic-antagonist drugs in the prevention of gastrointestinal bleeding in patients with cirrhosis and esophageal varices. An analysis of data and prognostic factors in 589 patients from four randomized clinical trials. N Engl J Med. 1991 May 30;324:1532-8. [PubMed ID: 1674104]



To determine the effectiveness of β-adrenergic antagonists (propranolol and nadolol) in preventing upper gastrointestinal hemorrhage in patients with cirrhosis and esophageal varices.

Data sources

Not stated.

Study Selection

Studies were selected if they were randomized controlled trials evaluating the use of β-adrenergic-antagonist drug treatment (propranolol and nadolol) in preventing a first episode of gastrointestinal bleeding caused by partial hypertension in patients with cirrhosis.

Data Extraction

4 randomized controlled trials were included in the meta-analysis. Individual patient data were collected from the physicians of the 589 patients that were studied. Patients had been randomized to propranolol (n = 203) or nadolol (n = 83) in doses that reduced the heart rate by approximately 25%, or to placebo (n = 165), ranitidine (n = 49), or vitamin K (n = 89). Fatal and nonfatal upper gastrointestinal bleeding and total mortality were reported over a mean follow-up period of 491 days.

Main results

17% of β-blocker patients had gastrointestinal bleeding compared with 27% of control patients. {This absolute risk reduction of 10% means that 9 patients would need to be treated (NNT) with β-blockers (compared with placebo) to prevent gastrointestinal bleeding in 1 additional patient, 95% CI 6 to 21; the relative risk reduction was 35%, CI 5% to 56%.}* Bleeding was more likely among patients with ascites, low serum albumin, and Child-Pugh score ≥ 8. When bleeding occurred, it was from varices in 71% and 61% of β-blocker and comparison patients, respectively. The case-fatality rates within 6 weeks of bleeding were 43% and 52% among β-blocker and comparison patients, respectively. Total mortality at 2 years was 29% among β-blocker patients and 29% among comparison patients (P > 0.2).


β-adrenergic-antagonist drugs are effective in reducing fatal and nonfatal first-time gastrointestinal bleeding in patients with cirrhosis and esophageal varices but do not improve total 2-year mortality.

Source of funding: Not stated.

Address for article reprint: Dr. T. Poynard, Hôpital Antoine Beclérè, F-92141 Clamart, France.

*Numbers calculated from data in article.


The high mortality of an initial variceal bleeding episode in a patient with cirrhosis and portal hypertension makes primary preventive treatment desirable. Results from individual randomized primary prevention trials of β-blockers are inconsistent, possibly because of differences in study populations; I believe such clinical heterogeneity limits the usefulness of a classical meta-analysis. Poynard and colleagues went a step further by combining the original data from 4 (I presume the most rigorous) of 7 published trials. Although their conclusions support the overall effectiveness of β-blockers in preventing a first upper gastrointestinal hemorrhage, questions concerning their clinical application remain. Which clinical variables predict response to β-blockers? Treatment was more effective among patients without ascites (RRR of 52%; NNT to prevent 1 bleeding episode = 9) and among patients with a Child-Pugh score ≥ 8 (RRR 49%; NNT = 7), but what about patients who satisfied both conditions? What about other variables such as serum albumin and prothrombin time? These questions could be addressed in this study but would require validation in a separate group of similar patients.

Perhaps the most important clinical issue concerns the generalizability of the conclusions of this overview. All study patients had evidence of cirrhosis, esophageal varices, and no previous upper gastrointestinal bleeding. Exclusions, which reduced eligible participants by approximately 25%, included contraindications to β-blockers (conditions that often co-exist with cirrhosis); hepatocellular carcinoma; intractable ascites or encephalopathy; and severe cirrhosis. Finally, withdrawals because of adverse effects of β-blocker therapy must be considered. In one trial (1), this accounted for 14% of randomized subjects.

This study answers some questions and generates several others. Before prophylactic nonselective β-adrenergic blockade can be recommended for all eligible patients with cirrhosis and esophageal varices, further research should be done to address these questions.

Thomas F. Imperiale, MD
Case Western Reserve UniversityCleveland, Ohio, USA


1. Italian Multicenter Project for Propranolol in Prevention of Bleeding. J Hepatol. 1989;9:75-83.

Updated Commentary

Since publication of the analysis by Poynard and colleagues 10 years ago, subsequent primary research and practice guidelines have established and supported the use of nonselective β-blockers (e.g., propranolol and nadolol) as first-line therapy for primary prophylaxis in patients with cirrhosis who have medium-to-large esophageal varices (1-5). After 25 months of β-blocker therapy, overall bleeding rates are 15% in treated patients compared with 25% in untreated patients (1). Dosing of β-blockers is aimed at reducing the resting heart rate by 25% or to a tolerated minimum of 55 beat/min.

As β-adrenergic blockade has become more widely accepted, emphasis has broadened to include identification of patients eligible for primary prophylaxis and establishment of predictors of large esophageal varices among patients with cirrhosis. Some guidelines have called for periodic endoscopy to screen for varices among patients with confirmed or suspected cirrhosis (1-3), although the cost-effectiveness of this practice remains to be determined. Several studies have identified clinical variables such as thrombocytopenia, splenomegaly, and portal vein diameter to help stratify risk for large esophageal varices and increase the efficiency of this practice (6).

What is less clear is the management of patients with contraindications or intolerance to β-blockers. Available evidence does not convincingly support the use of nitrates alone, and although endoscopic variceal ligation as primary prophylaxis appears promising, further study is needed. Additional important yet unresolved issues are whether β-blockers have any role in preventing the formation and growth of varices (preprimary prophylaxis), and identifying predictors of an optimal hemodynamic response to β-blocker therapy. Results of ongoing studies are anxiously awaited to address these unresolved issues.

1. D'Amico G, Pagliaro L, Bosch J. Pharmacological treatment of portal hypertension: an evidence-based approach. Semin Liver Dis. 1999;19:475-505.

2. Grace ND. Diagnosis and treatment of gastrointestinal bleeding secondary to portal hypertension. Am J Gastroenterol. 1997;92:1081-91.

3. Garcia-Tsao G. Current management of the complication of cirrhosis and portal hypertension: variceal hemorrhage, ascites, and spontaneous bacterial peritonitis. Gastroenterol. 2001;120:726-48.

4. de Franchis R. Updating consensus on portal hypertension: report of the Baveno III consensus workshop on definitions, methodology and therapeutic strategies in portal hypertension. J Hepatol. 2000;33:846-52.

5. Grace ND, Groszmann, Garcia-Tsao G, et al. Portal hypertensionand variceal bleeding: an AASLD single topic symposium. Hepatology. 1998;28:868-80.

6. Ong J. Clinical predictors of large esophageal varices: how accurate are they? Am J Gastroenterol. 1999;94:3103-5.