Applying effectiveness estimates from a network meta-analysis involving over 800,000 patients from randomised controlled trials and observation studies, our Markov model projected cost and health outcomes for a cohort of 65-year-old AF patients over a life-time. We performed subgroup analysis stratified by age (65–74 and ≥ 75 years), with further analysis limited to observational studies involving dabigatran and rivaroxaban.
Compared to warfarin, NOACs (except dabigatran 110) were associated with incremental cost-effectiveness ratios ranging from USD 24,476 to USD 41,448 that were within cost-effectiveness threshold of USD 49,700 (one gross domestic product per capita in Singapore in 2015). Aspirin regimens were dominated. In elderly aged ≥ 75 years, cost effectiveness of NOACs (except apixaban) decreased, owing to worsened performance in safety profile. Analysis limited to observational studies revealed that dabigatran 150 and rivaroxaban were not cost-effective, reflecting increased bleeding risks in non-controlled settings. Threshold analyses revealed that apixaban was no longer cost-effective at two to three times higher bleeding risk.
Whilst NOACs are cost-effective in the younger elderly compared to warfarin, their benefits appear to be offset by worsened risk profile in older elderly, especially in non-controlled settings. Decisions on appropriate AF treatment should balance treatment-related benefits, risks, and patient preference.