• ulcer rebleeding over period of 13 months
  

• stroke or cardio-vascular events
  

Among patients with previous ulcer bleeding, neither nonsteroidal anti-inflammatory drugs (NSAIDs) plus a proton pump inhibitor (PPI) nor a cyclooxygenase (COX)-2 inhibitor adequately prevents recurrent ulcer bleeding.

The aim is to investigate whether adding a PPI to a COX-2 inhibitor is superior to a COX-2 inhibitor alone for the prevention of recurrent ulcer bleeding in high-risk patients.

Patients with arthritis who presented with NSAID-induced ulcer bleeding were eligible if they had healed ulcers on follow-up endoscopy; were H. pylori negative or eradicated; and required regular NSAIDs. After the ulcers had healed, all patients received celecoxib 200 mg b.i.d. They were randomly assigned to esomeprazole 20 mg b.i.d. or its placebo for 12 months. The primary endpoint was recurrent ulcer bleeding.

Results Intention-to-treat analysis included 273 patients (137 received esomeprazole and 136 received placebo). Recurrent ulcer bleeding occurred in none receiving esomeprazole and 8 patients receiving placebo. The 12-month cumulative incidence of recurrent bleeding was 0% in the esomeprazole group and 8.9% in the placebo group (difference, 8.9 percentage points; 95% confidence interval for the difference, 4.1 to 13.7; P=0.0004). During follow-up, 16.1% of patients in the esomeprazole group and 15.4% in the placebo group used low-dose aspirin. Among patients who did not use aspirin, the cumulative incidence of recurrent ulcer bleeding was 0% in the esomeprazole group and 7.1% in the placebo group (difference, 7.1 percentage points; 95% confidence interval, 2.4 to 11.8; P=0.004).

Conclusion Among patients with arthritis who have previous NSAID-induced ulcer bleeding, adding a PPI to a COX-2 inhibitor is superior to a COX-2 inhibitor alone for the prevention of recurrent ulcer bleeding.