Randomized study comparing banding and propranolol to prevent initial variceal hemorrhage in cirrhotics with high-risk esophageal varices

Background & Aims: Standard care for prevention of first esophageal variceal hemorrhage is beta-blockade, but this may be ineffective or unsafe. Our purpose was to compare endoscopic banding with propranolol for prevention of first variceal hemorrhage. Methods: In a multicenter, prospective trial, 62 patients with cirrhosis with high-risk esophageal varices were randomized to propranolol (titrated to reducing resting pulse by 25%) or banding (performed monthly until varices were eradicated) and were followed up on the same schedule for a mean duration of 15 months. The primary end point was treatment failure, defined as the development of endoscopically documented variceal hemorrhage or a severe medical complication requiring discontinuation of therapy. Direct costs were estimated from Medicare reimbursements and fixed or variable charges for services up to treatment failure. Results: Background variables of the treatment groups were similar. The trial was stopped early after an interim analysis showed that the failure rate of propranolol was significantly higher than that of banding (6/31 vs. 0/31; difference, 1.9.4%; P =.0098; 95% confidence interval for true difference, 6.4%-37.2%). Significantly more propranolol than banding patients had esophageal variceal hemorrhage (4/31 vs. 0/31; difference, 12.9%; P =.0443; 95% confidence interval for true difference, 0.8%-29%), and the cumulative mortality rate was significantly higher in the propranolol than in the banding group (4/31 vs. 0/31; difference, 12.9%; P =.0443; 95% confidence interval for true difference, 0.8%-29%). Direct costs of care were not significantly different. Conclusions: For patients with cirrhosis with high-risk esophageal varices and no history of variceal hemorrhage, propranolol-treated patients had significantly higher failure rates of failure, first esophageal varix hemorrhage, and cumulative mortality than banding patients. Direct costs of medical care were not significantly different.