Cost-Effectiveness of Self-Expandable Transcatheter Aortic Valves in Intermediate-Risk Patients

A recent clinical trial showed that self-expandable transcatheter aortic valve replacement (TAVR) was non-inferior to surgical aortic valve replacement (SAVR). However, the cost-effectiveness of self-expandable TAVR in the intermediate-risk population remains unknown.

A cost-utility analysis from the Canadian health care system payer's perspective was undertaken to compare self-expandable TAVR with SAVR. A fully probabilistic Markov model over the patient's lifetime was constructed to estimate differences in costs (2016 Canadian dollars) and effectiveness (quality-adjusted life-years [QALYs]), discounted at 1.5% per annum. Incremental cost-effectiveness ratios (ICERs) were calculated. Efficacy inputs were obtained from the Surgical Replacement and Transcatheter Aortic Valve Implantation trial, and costs were primarily obtained from the Canadian Institute of Health Information. Probabilistic analysis (PA) and one-way deterministic sensitivity analyses were conducted around key point estimates to address uncertainty.

In the base case analysis, with discounting, the total lifetime costs (mean ± standard deviation) in the TAVR and SAVR arms were $44,299 ± $7,260 and $32,994 ± $13,434, respectively, whereas total effectiveness values were 6.42 ± 1.33 QALYs and 6.28 ± 1.32 QALYs, respectively. This yielded an incremental cost of $11,305 and incremental effectiveness of 0.15 QALYs when TAVR was compared with SAVR for an ICER of $76,736/QALY. In the PA, there was moderate uncertainty, with 52.8% and 57.2% of simulations less than willingness-to-pay thresholds of $50,000 and $100,000, respectively. In the sensitivity analysis, when the cost of TAVR valve system was priced at $17,397 (base case $22,000 Canadian dollars), TAVR was found to be cost-effective at a willingness-to-pay threshold of $50,000/QALY.

Self-expandable TAVR was found to be cost-effective; however, there was moderate uncertainty, reflecting the non-inferiority nature of the data.