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


Sclerotherapy for prophylaxis of hemorrhage from esophageal varices increased mortality

ACP J Club. 1992 May-June;116:72. doi:10.7326/ACPJC-1992-116-3-072

Source Citation

The PROVA Study Group. Prophylaxis of first hemorrhage from esophageal varices by sclerotherapy, propranolol or both in cirrhotic patients: a randomized multicenter trial. Hepatology. 1991 Dec;14:1016-24.



To evaluate sclerotherapy and propranolol, separately and combined, for prophylaxis of hemorrhage from esophageal varices in cirrhotic patients.


Randomized, placebo-controlled trial.


11 hospitals in Denmark and Norway.


Patients with biopsy-proven or clinically evident cirrhosis, with endoscopy-proven esophageal varices with no previous bleeding or sclerotherapy, for whom β-blocker therapy was not required or contraindicated and for whom repeat sclerotherapy was feasible, were allocated to receive sclerotherapy (n = 73), propranolol (n = 68), both (n = 73), or no therapy (n = 72).


Sclerotherapy was done with flexible endoscopes and polidocanol, maximum of 30 mL/session, at 1- to 2-week intervals until the varices were eradicated. This treatment was repeated for recurrences. Endoscopy was done every 3 months for the first year of follow-up, then every 6 months thereafter. Slow-release propranolol, 160 mg, was taken once daily and adjusted over 2 weeks in 80-mg increments until the heart rate was reduced by 25%. Maximum daily dose was 400 mg, minimum resting heart rate was 50 beats/min, and minimum systolic blood pressure was 90 mm Hg.

Main outcome measures

Death, or transfusion-requiring upper gastrointestinal bleeding probably caused by esophageal varices and occurring > 24 hours after sclerotherapy.

Main results

The trial was closed after 3 years because a preliminary analysis suggested a harmful effect of sclerotherapy and no detectable effect of propranolol. Mean length of follow-up was 15.4 months (range 0 to 41.5 months). The numbers of patients with ≥ 1 transfusion-requiring variceal hemorrhage in the sclerotherapy, propranolol, combined therapy, and control groups were 13 (18%), 12 (18%), 12 (16%), and 13 (18%), respectively. The numbers of deaths occurring after a bleeding episode were 5 (38%), 2 (17%), 8 (67%), and 6 (46%), respectively (P = 0.05 for the comparison between highest and lowest rates). Mortality rates with and without previous variceal bleeding were higher for the combined therapy than for the other therapies (P = 0.03 and P = 0.002, respectively).


Sclerotherapy for prevention of hemorrhage from esophageal varices in cirrhotic patients was associated with an excess mortality rate; propranolol alone was ineffective.

Sources of funding: Danish Medical Research Council; ICI Pharmaceuticals Inc.; Kreussler Inc.

Address for article reprint: Dr. T.I. Sørensen, Department of Medicine 261, Hvidovre Hospital, DK-2650 Copenhagen, Denmark.


This important, carefully planned and conducted multicenter study of prophylactic treatment of esophageal varices in patients with cirrhosis is the largest study of its kind. It shows clearly that sclerotherapy not only does not prevent the first hemorrhage from varices but is actually detrimental in this patient population. Sclerotherapy significantly increased mortality. The results of this study confirm the results of another large, carefully conducted study of prophylactic sclerotherapy done by the Veterans Affairs Cooperative Group (1). What is striking and probably not coincidental is that both studies were stopped early because of excess mortality in the group of patients receiving sclerotherapy. In both studies no single cause of death existed. The message, however, is clear—repeated sclerotherapy sessions should not be done because they are poorly tolerated by patients with cirrhosis who have not yet bled from varices.

An additional result of this study, that propranolol did not decrease the likelihood of a first variceal hemorrhage, is at variance with the positive results of other studies of nonspecific β-blockers. The reasons for this discrepancy are unclear, but the differences were small. No study of β-blockers has clearly shown a deleterious effect of prophylactic β-blockage, and meta-analysis of previously published studies, including this one, suggests that β-blockers do decrease the likelihood of an initial variceal hemorrhage by approximately 50% (2). Hence, in contrast to prophylactic sclerotherapy, propranolol or nadolol may still be used by those who favor a more aggressive approach in patients who have not yet bled from varices.

Marshall M. Kaplan, MD
Tufts UniversityBoston, Massachusetts, USA

Marshall M. Kaplan, MD
Tufts University
Boston, Massachusetts, USA


1. The Veterans Affairs Cooperative Variceal Sclerotherapy Group. Prophylactic sclerotherapy for esophageal varices in men with alcoholic liver disease. N Engl J Med. 1991;324:1779-84.

2. Pagliaro L, Burroughs AK, Sørensen TI, et al. Beta-blockers for preventing variceal bleeding. Lancet. 1990;336:1001-2.