The relationship between Cox-2 inhibitors and cardiovascular risk: a retrospective analysis using the Veteran Affairs (VA) database

Access full-text files




Motsko, Stephen Paul

Journal Title

Journal ISSN

Volume Title



The search for non-steroidal anti-inflammatory drugs (NSAIDs) with less gastrointestinal toxicity, led to the introduction of the selective cyclooxygenase-2 (COX-2) inhibitors. However, with this introduction into the market, there have been concerns regarding their safety, particularly cardiovascular safety.

The purpose of this study was to assess the cardiovascular risk (events included: myocardial infarction, stroke, and myocardial infarction-related deaths) associated with long-term (after 180 days of exposure) and short-term (≤ 180 days exposure) use of nonselective NSAIDs and COX-2 inhibitors. A retrospective analysis of the Veterans Integrated Service Network 17 Veteran’s Affairs (VA) database was conducted. Medicare data and Texas Department of Health mortality data were incorporated to capture events occurring outside the VA healthcare network. vii Patients 35 years of age and older receiving celecoxib, rofecoxib, ibuprofen, etodolac, and naproxen from January 1, 1999 through December 31, 2001, were included. Multivariate Cox proportional hazard models were used to analyze the relationship between cardiovascular risk and NSAID/COX-2 inhibitor use while adjusting for various risk factors. We identified 12,194 exposure periods and 146 cardiovascular events over the entire study period. Compared to long-term ibuprofen use, long-term use of celecoxib (Celebrex®) was associated with a 3.64 fold (95% CI 1.36 – 9.70; p = 0.01) increase in cardiovascular risk. Long-term use of rofecoxib (Vioxx ®) was associated with a 6.64 fold (95% CI 2.17 – 20.28; p ≤ 0.01) increased risk when compared to long-term use of ibuprofen. Short-term use of celecoxib and rofecoxib was not associated with any significant change in cardiovascular risk when compared to ibuprofen use (celecoxib – adjusted risk ratio (aRR) 0.67; 95% CI 0.31 – 1.47; p = 0.32 / rofecoxib – aRR 1.40; 95% CI 0.59 – 3.33; p = 0.44). Long-term and short-term exposure to naproxen and etodolac was not associated with a cardionegative or cardioprotective effect when compared to ibuprofen use. Additionally, overall exposure (long-term and short-term exposure) to study medications was evaluated; results indicated no significant findings with any COX2 inhibitor/NSAID. viii The findings of this observational study along with recent clinical trial results suggest that long-term exposure to COX-2 inhibitors is associated with an increased cardiovascular risk. In addition, the study results do not support the hypothesis that naproxen provides cardioprotective effects.