MEDLINE, EMBASE and the Cochrane Library (from their inception to May 2015)were searched for randomised controlled trials comparing the risk of gastrointestinaladverse events in patients taking nonselective NSAID s, selecti ve cyclooxy genase(COX)-2 inhib itors or nonselectiv e NSAIDs/COX-2 inhib itors plus gastroprotecti veagents [proton pump inhibitors (PPIs), histamine-2 receptor antagonists, misopros-tol]. Both pairwise meta-analysis and Bayesian network meta-analysis were per-formed.