It is important to understand differences between dual cyclooxygenase-inhibiting NSAIDs and coxibs for effective use in appropriate populations. The ADVANTAGE (Assessment of Differences between Vioxx And Naproxen To Ascertain Gastrointestinal tolerability and Effectiveness) study was the largest trial to enroll patients with osteoarthritis using less restrictive entry criteria. In addition, it permitted use of low-dose aspirin during the trial and involved a sample that was characteristic of patients seen in clinical practice with typical comorbid conditions. Treatment of osteoarthritis with rofecoxib and naproxen in this setting was designed to study "real-world" conditions. Other important coxib studies, such as the VIGOR study, which included patients with rheumatoid arthritis who were not taking low-dose aspirin, and CLASS (Celecoxib Long-term Arthritis Safety Study), which included both patients with osteoarthritis and those with rheumatoid arthritis and permitted low-dose aspirin, had differences in design (1, 2).

Our study design compared the highest doses of rofecoxib and naproxen indicated for long-term use in patients with osteoarthritis, according to prescribing information, to fully assess potential differences in tolerability. As pointed out by Dr. Hanauer, a higher dose of rofecoxib, 50 mg, is available; however, it is not recommended for long-term use. The main benefit of coxibs compared with NSAIDs relates to GI side effects, a major cause of drug-induced morbidity and mortality. Our study and VIGOR (1) demonstrated improvement in general GI tolerability and reductions in serious GI events (perforations, ulcers, GI bleeding) with rofecoxib compared with naproxen. As noted by Drs. Rusche and Jenkinson, rates of discontinuation due to causes other than GI events, which were not prespecified or powered for in our study, may be similar with these medications. The meta-analysis by Konstam and colleagues, which compared thrombotic event rates in rofecoxib studies, included data from our study (3). The report by Konstam and colleagues and our study both used the end point defined by the Antiplatelet Trialists' Collaboration, a standard for this type of safety reporting. Although the rate was lower with naproxen, the incidence of Antiplatelet Trialists' Collaboration events was similar with rofecoxib and placebo in the meta-analysis. The relative risk for rofecoxib compared with placebo was 0.84 (95% CI, 0.51 to 1.38), providing no evidence of excess cardiovascular events (3). In our study, 13% of patients used low-dose aspirin. Although it has been shown that rofecoxib does not alter platelet inhibition induced by low-dose aspirin (4), we had insufficient data to evaluate this clinically.

Coxibs can reduce GI adverse events and may reduce costs (5, 6). Further study is needed to determine whether differences between coxibs and NSAIDs for non-GI end points affect cost-effectiveness. It is ultimately the practitioner's responsibility to identify patients specifically at risk for drug-associated side effects and to appropriately select therapy using an evidence-based approach (5).