To the Editors:
We thank Mr. Loyd and colleagues for their thorough review of our study. Mr. Loyd et al have several points of contention, which we will discuss sequentially.
First, Mr. Loyd and colleagues contend that our model is limited because it includes data for both celecoxib and now withdrawn rofecoxib. We included data for both coxibs because our study was conducted prior to rofecoxib's removal from the market. Moreover, we explicitly biased our model in favor of coxibs by testing a hypothetical “best-case” coxib that represents the most favorable hybrid between celecoxib and rofecoxib. Four estimates exemplify this bias. First, rather than model a higher rate of ulcer complications for rofecoxib than for celecoxib, as supported by published data (1), we assumed a favorable 50% risk reduction in ulcer complications for both coxibs compared with nonselective NSAIDs. Second, rather than assume that rofecoxib provided no risk reduction in dyspepsia after 6 months, as supported by published data (2), we assumed that both coxibs provided a 23% reduction versus nonselective NSAIDs over the course of the entire time horizon. Third, rather than assume that the efficacy of celecoxib was inferior to that of rofecoxib, as supported by randomized controlled data in OA (3), we assumed that both coxibs were equally effective in providing symptom relief for arthritis pain. Finally, rather than estimate a daily average consumption of 1.4 celecoxib pills and 1.1 rofecoxib pills, as supported by administrative claims data (4), we assumed that only 1 pill was required daily for all coxibs. Therefore, where clinical data tend to disfavor rofecoxib (i.e., upper-GI hemorrhage and dyspepsia rates), we used celecoxib data, and where clinical data tend to disfavor celecoxib (i.e., arthritis efficacy and daily average consumption), we adopted rofecoxib data. Despite modeling this “best-case” hybrid coxib, our analysis suggests that coxibs may not be cost-effective.
Second, Mr. Loyd et al point out that our study included data from trials involving both RA and OA and that “celecoxib is mainly used to treat OA,” and in our opinion, this criticism lacks weight not only because coxibs are indeed used for both OA and RA, but also because there is no a priori reason to expect that the cost-ineffectiveness of coxibs would vary significantly by indication. The fact that coxibs were dominated in our analysis (i.e., more expensive yet less effective than alternatives) is unlikely to be overcome by subdividing data by type of arthritis.
Third, Mr. Loyd et al argue that our assumption that ASA blunts the GI safety of coxibs is not supported by robust evidence and does not comply with data from the SUCCESS study. However, our assumption is derived precisely from robust clinical evidence, including the SUCCESS study (reference 10 in our manuscript). Moreover, the largest published randomized controlled trial to date reporting clinically significant ulcer complications in patients receiving coxibs plus ASA versus NSAID plus ASA (CLASS study) (5) revealed no significant differences in complicated GI events between interventions. An error in our table (that is correct in the text itself) indicates that our assumption is based on a previously published decision analysis, that is incorrect. Mr. Loyd et al mistakenly seize on this error in their letter, but it should be quite evident that we could not have based a point estimate on a previous decision analysis, which itself does not contain any primary data. In short, there is little debate that the relative GI safety of coxibs is probably undermined by ASA. But despite this clarity, we nonetheless performed a sensitivity analysis in which we assumed no impact of ASA on relative GI safety of coxibs, and still found that coxibs are not cost-effective.
Fourth, Mr. Loyd and colleagues present a series of interlocking statements which, taken together, argue that our assumption that coxibs might promote cardiovascular adverse events is overstated. Specifically, they claim that celecoxib does not raise the risk of serious cardiovascular thrombotic events. This may be correct, but the black box warning on celecoxib issued by the Food and Drug Administration argues otherwise. More to the point, we conducted a sensitivity analysis in which we assumed no difference in cardiovascular safety between arms, and still found that coxibs are not cost-effective.