PURPOSE: To assess the risk of gastrointestinal perforation, ulcers, or bleeding (PUB) associated with the use of conventional nonsteroidal anti-inflammatory drugs (NSAIDs) with proton pump inhibitors (PPIs) and selective COX-2 inhibitors, with orithout PPIs compared with conventional NSAIDs. METHODS: A case-control study was performed within conventional NSAIDs and/or selective COX-2 inhibitors users identified from the Dutch PHARMO Record Linkage System in the period 1998-2012. Cases were patients aged >/=18 years with a first hospital admission for PUB. For each case, up to four controls were matched for age and sex at the date a case was hospitalized (index date). Logistic regression analysis was used to calculate odds ratios (ORs). RESULTS:At the index date, 2634 cases and 5074 controls were current users of conventional NSAIDs or selective COX-2 inhibitors. Compared with conventional NSAIDs, selective COX-2 inhibitors with PPIs had the lowest risk of PUB (adjusted OR 0.51, 95% confidenceinterval [CI]: 0.35-0.73) followed by selective COX-2 inhibitors (adjusted OR 0.66, 95%CI: 0.48-0.89) and conventional NSAIDs with PPIs (adjusted OR 0.79, 95%CI: 0.68-0.92). Compared with conventional NSAIDs, the risk of PUB was lower for those aged >/=75 years taking conventional NSAIDs with PPIs compared with younger patients (adjusted interaction OR 0.79, 95%CI: 0.64-0.99). However, those aged >/=75 years taking selective COX-2 inhibitors, the risk was higher compared with younger patients (adjustedinteraction OR 1.22, 95%CI: 1.01-1.47). CONCLUSIONS: Selective COX-2 inhibitors with PPIs, selective COX-2 inhibitors, and conventional NSAIDs with PPIs were associated with lower risks of PUB compared with conventional NSAIDs. These effects were modified by age. (c) 2017 The Authors. Pharmacoepidemiology & Drug Safety Published by John Wiley & Sons Ltd.