The results were reported separately for all patients, medium-to-high risk patients, and high-risk patients, assuming a cohort of 1,000 treated patients in each group.
For all patients, the number of deaths due to serious gastrointestinal adverse events ranged from 0.8 with NSAIDs alone to 0.31 with celecoxib. The expected costs ranged from EUR 166,000 with NSAIDs alone to EUR 284,000 with NSAIDs plus H2RAs, and EUR 285,000 with NSAIDs plus PPIs.
NSAIDs plus H2RAs and NSAIDs plus misoprostol were dominated by Arthrotec as it was more effective and cheaper. NSAIDs plus PPIs was dominated by celecoxib. These dominated treatments were excluded from the incremental analysis.
For all patients, NSAIDs alone resulted in an incremental cost-effectiveness ratio (ICER) of EUR 17 per LYG. Arthrotec had an ICER of EUR 5,676 per LYG compared with NSAIDs alone. Celecoxib had an ICER of EUR 56,667 compared with Arthrotec.
For medium-to-high risk patients excluding dominated treatments, NSAIDs alone resulted in an ICER of EUR 20 per LYG. Arthrotec compared with NSAIDs alone resulted in an ICER of EUR 526 per LYG and celecoxib compared with Arthrotec resulted in an ICER of EUR 33,684 per LYG.
For high risk patients, NSAIDs alone were also dominated, Arthrotec resulted in an ICER of EUR 23 and celecoxib compared with Arthrotec resulted in an ICER of EUR 15,429 per LYG.
The probabilistic sensitivity analysis showed that at a willingness-to-pay threshold of EUR 20,00 per LYG , for all patients, NSAIDs alone had a 44% probability of being the most cost-effective strategy, Arthrotec had a 38% probability and celecoxib 15%. Conversely, for high-risk patients, Arthrotec had a 41% probability of being the most cost-effective strategy and celecoxib 48%.