Risk factors for upper gastrointestinal bleeding in patients taking selective COX-2 inhibitors: A nationwide population-based cohort study

Abstract

Objective. Cyclooxygenase-2 inhibitors (coxibs) are associated with less upper gastrointestinal bleeding (UGIB) than traditional nonsteroidal antiinflammatory drugs (tNSAIDs). However, they also increase the risk of UGIB in high-risk patients. We aimed to identify the risk factors of UGIB in coxibs users. Design. Retrospective cohort study. Setting. 2000-2010 National Health Insurance Research Database of Taiwan. Subjects. Patients taking coxibs as the study group and patients not taking any coxibs as controls. Methods. After age, gender, and comorbidity matching by propensity score, 12,145 coxibs users and 12,145 matched controls were extracted for analysis. The primary end point was the occurrence of UGIB. Cox multivariate proportional hazard regression models were used to determine the risk factors for UGIB among all the enrollees and coxibs users. Results. During a mean follow-up of three years, coxibs users had significantly higher incidence of UGIB than matched controls (P < 0.001, log-rank test). Cox regression analysis showed that coxibs increased risk of UGIB in all participants (hazard ratio 5 1.37, 95% confidence interval 5 1.19-1.55, P < 0.001). Independent risk factors for UGIB among coxibs users were age, male gender, diabetes, chronic renal disease, cirrhosis, history of peptic ulcer disease, PU bleeding (PUB), Helicobacter pylori (H. pylori) infection, and concomitant use of tNSAIDs, acetylsalicylic acid, or thienopyridines. Conclusions. Among coxibs users, H. pylori infection and history of PUB were especially important risk factors for UGIB. Further studies are needed to determine whether proton pump inhibitors might play a protective role in these at-risk patients.