Thirty-two trials (n=72,584) were included in the review. Two of the trials were analysed in more than one outcome category. The majority of the trials were randomised and controlled, but the specific design of those addressing risk of CVEs was not given (8 studies).
Three (n=5,838) of the 22 trials examining this outcome showed coxibs to be significantly more effective for pain control than non-selective NSAIDS. Two of the trials compared rofecoxib with nabumetone and showed statistically significant improvements in response rates (p<0.003) and mean walking pain (p=0.031). The other trial comparing etoricoxib with naproxen demonstrated improvements in pain control and tolerability (p<0.05). The remaining 19 trials showed comparable or non significantly different results.
Three trials examined the risk of GIs, of which one (n=287) directly compared the upper GI event rate of celecoxib with a combination treatment of non-selective NSAID (diclofenac) and PPI (omeprazole). At the 6-month follow-up, there were no significant differences between the treatments for recurrent ulcer bleeding or endoscopically-determined ulcers. Another trial (n=356) compared celecoxib with naproxen plus omeprazole, and with placebo. The incidence of lower GI (small bowel) events was significantly lower amongst the coxib treatment group (p<0.001). This trend was confirmed in a further trial (n=8,076) of rofecoxib compared with naproxen, but without a PPI (p=0.032).
Trials comparing coxibs with placebo provided the strongest evidence that coxibs may significantly increase the risk of CVEs. Unpublished data suggested that the risk may be dose-related. Of the 8 trials in this outcome category, the results of one (n=1,671) (in which CVEs were used as the primary measure) suggested a statistically significantly elevated risk of CVE following parecoxib and valdecoxib compared with placebo (RR 3.7, 95% CI: 1.0, 13.5). This trend was demonstrated in another 2 trials. The first (n=2,586) compared rofecoxib with placebo and found a two-fold increase in risk (RR 1.92, 95% CI: 1.19, 3.11). The second (n=2,035) found a three-fold increase in risk (RR 3.4, 95% CI: 1.4, 7.8) when a higher dose of celecoxib was compared with placebo.
The evidence was less strong when coxibs were compared with non-selective NSAIDS. One trial (n=8,076) found a four-fold elevated risk of mycardial infarction associated with rofecoxib (0.4% versus 0.1% with naproxen) but the difference was not significant. Another trial (n=8,059) found no statistically significant increase associated with celecoxib when compared with ibuprofen or diclofenac. A larger trial (n=18,325) also found no significant increase in CVEs when lumiracoxib was compared with naproxen or ibuprofen.