Cost-Effectiveness Analysis of Linagliptin in Japan Based on Results from the Asian Subpopulation in the CARMELINA® Trial

Abstract

Introduction: We evaluated the cost-effectiveness of linagliptin in Japan by estimating the lifetime outcome based on clinical event rates from the Asian subpopulation of the CARMELINA trial. In CARMELINA, linagliptin added to standard of care (SoC) versus SoC demonstrated noninferiority with regard to risk of composite cardiovascular (CV) outcome in patients with type 2 diabetes at high risk of CV and kidney events. Issues resulting from conducting a cost-effectiveness analysis using data from a clinical noninferiority study were also investigated. Methods: A microsimulation model was used to evaluate linagliptin/SoC versus SoC in terms of direct costs and quality-adjusted life years (QALYs) from a Japanese public healthcare payer’s perspective. Cost data were obtained from recent Japanese publications. The time horizon was defined as lifetime, and the discount rate for costs and effectiveness was 2% per year. One-way and probabilistic sensitivity analyses were performed. Results: In the base case analysis, and taking medical history into account, the incremental effectiveness of linagliptin/SoC versus SoC was 1.34 QALYs, and the incremental cost for linagliptin was − 545,319 yen. In the one-way sensitivity analysis, the parameter which most affected the results was the hazard ratio for renal failure of linagliptin/SoC compared with SoC. The probabilistic sensitivity analysis showed that the probability of reduced costs and increased effectiveness (dominant) was 48%. Assuming an incremental cost-effectiveness ratio (ICER) threshold of 5 million yen, the probability that the ICER was below the threshold was 89% for linagliptin/SoC compared with SoC. Conclusions: This evaluation, using Asian subpopulation data from the CARMELINA trial, suggested that the cost-effectiveness of linagliptin for a lifetime outcome was favourable in Japan. However, the results must be interpreted cautiously because of the noninferiority trial data source, which might cause ICER variations for each parameter.