Because dyspeptic symptoms are far more prevalent than ulcer complications in users of nonsteroidal anti-inflammatory drugs (NSAIDs), economic models indicate that dyspepsia rates (not ulcer complications) are the major determinant of cost-effectiveness in treating arthritis. We performed a meta-analysis to compare rates of dyspepsia for two common therapies in high-risk patients with arthritis: cyclooxygenase-2 inhibitor (Coxib) alone and combination therapy with a nonselective NSAID and a proton pump inhibitor (PPI) (NSAID+PPI).

We performed a systematic review to identify trials comparing either a Coxib versus NSAID or NSAID+PPI versus NSAID in chronic arthritis. We selected studies that report incident dyspepsia, defined a priori as “epigastric pain,” “dyspepsia,” and “nausea.” We then performed meta-analysis to compare the relative risk reduction and absolute risk reduction of dyspepsia for Coxib versus NSAID and NSAID+PPI versus NSAID.

Meta-analysis of 26 studies comparing dyspepsia between Coxibs and NSAIDs revealed a 12% relative risk reduction for Coxibs with an absolute risk reduction of 3.7%. Meta-analysis of four studies comparing dyspepsia between the NSAID+PPI combination and NSAIDs alone revealed a 66% relative risk reduction for NSAID+PPI with an absolute risk reduction of 9%. Compared with the NSAID strategy, the number needed to treat to prevent dyspepsia was 27 for Coxibs and 11 for NSAID+PPI.

NSAID+PPI affords greater risk reduction for dyspepsia than Coxibs when compared with the common baseline of NSAIDs. Because there are limited head-to-head data comparing Coxibs versus NSAID+PPI, these data provide the best indirect evidence that NSAID+PPI may be superior to Coxibs in minimizing incident dyspepsia.