Cost-effectiveness analysis of transcatheter aortic valve implantation in aortic stenosis patients at low- and intermediate-surgical risk in Japan

Abstract

Objective

To analyze the cost-effectiveness of transcatheter aortic valve implantation (TAVI) using the SAPIEN 3 (Edwards Lifesciences, Irvine, CA) compared to surgical aortic valve replacement (SAVR) in low- and intermediate-risk patients from a Japanese public healthcare payer perspective.

Methods

A Markov model cost-effectiveness analysis was developed. Clinical and utility data were extracted from a systematic literature review. Cost inputs were obtained from analysis of the Medical Data Vision claims database and supplemented with a targeted literature search. The robustness of the results was assessed using sensitivity analyses. Scenario analyses were performed to determine the impact of lower mean age (77.5 years) and the effect of two different long-term mortality hazard ratios (TAVI versus SAVR: 0.9–1.09) on both risk-level populations. This analysis was conducted according to the guidelines for cost-effectiveness evaluation in Japan from Core 2 Health.

Results

In intermediate-risk patients, TAVI was a dominant procedure (TAVI had lower cost and higher effectiveness). In low-risk patients, the incremental cost effectiveness ratio (ICER) for TAVI was ¥750,417/quality-adjusted-life-years (QALY), which was below the cost-effectiveness threshold of ¥5 million/QALY. The ICER for TAVI was robust to all tested sensitivity and scenario analyses.

Conclusions

TAVI was dominant and cost-effective compared to SAVR in intermediate- and low-risk patients, respectively. These results suggest that TAVI can provide meaningful value to Japanese patients relative to SAVR, at a reasonable incremental cost for patients at low surgical risk and potentially resulting in cost-savings in patients at intermediate surgical risk.

Graphical abstract

Cost-effectiveness analysis of transcatheter aortic valve implantation in aortic stenosis patients at low- and intermediate-surgical risk in Japan. Results demonstrate that TAVI can provide meaningful value to Japanese patients with severe symptomatic aortic stenosis relative to SAVR, at a reasonable incremental cost for low-risk patients and potentially resulting in cost-savings in intermediate-risk patients.

PLAIN LANGUAGE SUMMARY

Aortic Stenosis (AS) is the most common valvular heart disease in Japan, and, if left untreated, severe symptomatic AS (sSAS) is associated with a dramatic increase in mortality and morbidity. Transcatheter Aortic Valve Implantation (TAVI) is a minimally invasive treatment option for replacing the aortic valve in patients with sSAS and has been associated with similar or better outcomes compared to Surgical Aortic Valve Replacement (SAVR), which involves open-heart surgical replacement of the aortic valve. The objective of this study was to compare the costs and health outcomes associated with TAVI compared to SAVR in Japanese patients deemed low- or intermediate-risk for surgery. Despite the expanding use of TAVI in Japan, a cost-effectiveness analysis (CEA) does not exist that evaluates the economics of TAVI with the current generation SAPIEN 3 implant in patients with low- and intermediate-risk from a public perspective. Our study suggests that TAVI represents strong value for money among low- and intermediate-risk patients in Japan: compared to SAVR, TAVI is associated with better clinical outcomes and quality of life for patients, at a reasonable additional cost for low-risk patients and at a lower cost for intermediate-risk patients.

Keywords:

• Aortic stenosis
• cost utility analysis
• transcatheter aortic valve replacement
• incremental cost-effectiveness ratio
• Japan

JEL CLASSIFICATION CODES:

• I10
• I1
• I
• I13
• I00

Introduction

Aortic stenosis (AS) is the most common type of valvular heart disease in high-income countries and its prevalence increases with age.^1,2 If left untreated, severe symptomatic AS (ssAS) is associated with a dramatic increase in morbidity and mortality^3,4 and the progression of AS leads to increasing healthcare costs.⁵ A recent study of valvular heart disease in Japan found that less than 10% of patients with AS underwent valve repair or replacement,⁶ despite studies showing that valve replacement is associated with a substantial decrease in mortality and heart failure hospitalizations at 2 years in the Japanese population.⁷

Transcatheter aortic valve implantation (TAVI) is an innovative treatment for ssAS, exhibiting short- and mid-term outcomes that are comparable or even superior to surgical aortic valve replacement (SAVR) regardless of surgical risk level.^8–12 The use of TAVI has dramatically increased in Japan since it was first performed in 2009, with a 740% increase in the number of TAVI cases from 2014 to 2019 and more than 170 Japanese centers performing TAVI in 2019.¹³ Most recently, in March 2021, the Ministry of Health, Labor, and Welfare approved the treatment of patients with low surgical risk in Japan with SAPIEN 3 valves (Edwards Lifesciences, Irvine, CA) and gave reimbursement approval in August of the same year.¹⁴ As such, it is expected that the utilization of TAVI may grow even further.

Given the rapidly aging population of Japan and the corresponding escalation of healthcare costs, the economic evaluation of new medical technologies – including transcatheter aortic valves – is increasingly important.¹⁵ Despite the ever-expanding clinical implementation of TAVI in Japan, a cost-effectiveness analysis (CEA) does not exist that evaluates the economics of TAVI with the current generation SAPIEN 3 implant in patients with low and intermediate surgical risk from a public perspective. To address this gap in knowledge, we performed a CEA of TAVI with SAPIEN 3 in Japan in low- and intermediate-risk patients with ssAS from the perspective of the healthcare system in accordance with the guidelines for cost-effectiveness evaluation in Japan from Core 2 Health (C2H).¹⁶

Methods

Study overview and population

A CEA was developed to evaluate two treatment options, namely TAVI with SAPIEN 3 compared to SAVR. To assess clinical benefit, the C2H guideline recommendations were followed and quality-adjusted life-years (QALYs) were utilized.¹⁶ C2H is a department of the national institute of public health in Japan established to provide guidance for academic research on cost-effectiveness evaluation.¹⁶ The analyses were conducted from a Japanese public healthcare payer perspective utilizing a lifetime horizon and applying a 2% discount rate to both costs and QALYs.¹⁶ Populations of interest were ssAS patients at low and intermediate surgical risk, the definitions of which were based on indications of TAVI and risk levels described in published clinical trials.^8–10 Baseline patient age was set to an average age of 83 years as this is the average age of patients who underwent TAVI in Japan between 2019 and 2022, according to the Medical Data Vision (MDV) database. The MDV database is a large-scale Japanese administrative healthcare claims database derived from 480 hospitals accredited as providing acute care services in Japan, comprising approximately 26% of advanced treatment hospitals with information on the status and treatment of approximately 44 million patients.¹⁷

Model framework

A literature-based cost-effectiveness model was employed to estimate direct healthcare costs and quality-adjusted survival for TAVI with SAPIEN 3 compared to SAVR. All patients started the model with procedure: either TAVI or SAVR, and early adverse events for TAVI or SAVR were captured in the procedure state using a decision tree. After 30 days, patients entered a Markov model and transitioned to the long-term health states. The Markov model was developed with specific transition probabilities for a given patient based on their selected treatment, with a monthly cycle for simulation of the first 5 years followed by 1-year cycles until all patients exited the model (lifetime horizon), commencing at the initial TAVI or SAVR procedure (Figure 1). Transition between health states occurred through 1 year in the intermediate risk analysis and through 5 years in the low risk analysis, based on the duration of follow-up for each clinical study. Beyond the follow-up period, we made the conservative assumption that long-term clinical event rates were equivalent between TAVI and SAVR in the base case. In the base case, all-cause mortality after 1 year in the intermediate risk model and 5 years in the low-risk model was determined from the age-matched mortality of the general population in Japan.

Figure 1. Markov model based on complications observed in patients with ssAS. Abbreviations. PVL, Paravalvular leakage; SAVR, Surgical aortic valve replacement; ssAS, Severe Symptomatic Aortic Stenosis; TAVI, Transcatheter Aortic Valve Implantation.

For the cost analysis, procedural complications were defined using International Classification of Disease (ICD)-10 codes and Kubun codes (Supplemental Table 1). The complication event rates were based on 30-day clinical outcomes from the PARTNER 3 and PARTNER 2 S3i trials.^9–12 The Markov model had the following post-treatment health states: dialysis, moderate/severe paravalvular leakage (PVL), disabling stroke, and death. Transition probabilities were derived from the PARTNER 2 S3i trial for intermediate-risk patients and the PARTNER 3 trial for low-risk patients. Patients exited the model by moving from any health state to death.

Model parameters

The clinical and utility outcomes for TAVI with SAPIEN 3 and SAVR were obtained through two systematic literature reviews (SLR) (see Supplemental Methods and Supplemental Figures 1 and 2 for details). For patients through 30 days post-treatment, the clinical outcomes were based on 30-day outcomes from low- and intermediate-risk patients from PARTNER 3 and PARTNER 2 S3i trials, respectively (Table 1). For the intermediate risk population, monthly transition probabilities for complications that occurred after 30 days were obtained from the 1-year data for PARTNER 2 S3i. Because 5 year data for the PARTNER 3 trial was recently published, the 5-year data for PARTNER 3 was used to obtain transition probabilities after 30 days in the low risk population.²³ The probability of death resulting from stroke, dialysis, and moderate/severe PVL as hazard ratios was derived from an analysis of the PARTNER 2 studies (Supplemental Table 2).²⁴ Age-based mortality rates after the clinical trial period (1 year for intermediate-risk and 5 years for low-risk) were obtained from Japan life tables (Supplemental Table 3).²⁵

Table 1. Clinical event probabilities and utilities used in the model.

	Intermediate Risk			Low Risk
Clinical events	TAVI with SAPIEN 3	SAVR	Source	TAVI with SAPIEN 3	SAVR	Source
Probabilities of peri-operative complications at 30 days
Death	1.1%	4.0%	PARTNER II S3i^12	0.4%	1.1%	PARTNER 3 trial^10
Stroke	2.7%	6.1%		0.6%	2.4%
New pacemaker implantation	10.2%	7.3%		6.5%	4.0%
Kidney failure	0.5%	3.3%		0.4%	1.8%
Major bleeding	4.6%	46.7%		3.6%	24.5%
Myocardial infarction	0.3%	1.9%		1.0%	1.3%
Major vascular complications	6.1%	5.4%		2.2%	1.5%
Atrial fibrillation	5.0%	28.3%		5.0%	39.5%
Rehospitalization	4.6%	6.8%		3.4%	6.5%
Probabilities of post-operative complications at 1 year for intermediate-risk patients and 2 years for low-risk patients
Death	7.4%	13.0%	PARTNER II S3i^9^,^12	2.4%	3.2%	PARTNER 3 trial^10^,^11
Disabling stroke	2.3%	5.9%		0.8%	1.1%
Dialysis	2.2%	5.2%		0.2%	0.7%
Moderate/severe PVL	1.5%	0.4%		0.5%	0%
Rehospitalization	11.4%	15.1%		8.5%	12.5%
Utilities
Prior to intervention	0.733	0.727	^18^,^19	0.769	0.770	^19^,^20
Discharge to 6 months post-AVR	0.799	0.738		0.847	0.758
Post-AVR, 6 months–1 year	0.799	0.738		0.840	0.841
Post-AVR, >1 year	0.791	0.785		0.825	0.832
Disabling stroke	−0.265	−0.265	^21	−0.265	−0.265	^21
Dialysis	0.616	0.616	^22	0.744	0.744	^22

Abbreviations. PVL: Paravalvular Leakage; SAVR: Surgical Aortic Valve Replacement; TAVI: Transcatheter Aortic Valve Implantation.

The EQ-5D results from PARTNER 3 and PARTNER 2 S3i were used to determine the treatment utility weights (Table 1), converting to Japanese utilities using the value set from Shiroiwa et al.,²⁶ consistent with previous studies.^27,28 For long-term disabling stroke and dialysis, the utility decrements for the Japanese population were obtained from a literature search similar to that done for clinical inputs (Supplemental Table 4). Moderate/severe PVL was not assumed to impact utility, since no utility decrements or values were available for patients with this health state.²¹

Costs encompassed two categories: 1) Costs incurred in the hospital including device cost, procedure fee, and expenses associated with complications and 2) postoperative costs including long-term follow-up costs associated with disabling stroke and dialysis, as well as cost of rehospitalization (Table 2).

Table 2. Breakdown of TAVI and SAVR cost.

	TAVI with SAPIEN 3	Source	SAVR	Source
Peri-operative costs
Device cost	¥4,510,000 (code: 710010745)	^29	¥984,000	Claims data analysis^a
Procedure fee	¥390,600 (code: K555-2)	^30	¥855,000 (code: K555)	^30
Incremental cost of procedural complications
Other hospitalization cost^b	¥1,234,828	Regression analysis^c	¥3,554,266	Regression analysis^c
In-hospital mortality	¥1,516,442		¥2,474,105
Stroke	¥1,121,531		¥1,894,908
Pacemaker implantation	¥1,566,299		¥1,841,697
Kidney failure	¥1,823,416		¥1,058,893
Myocardial infarction	¥801,742		¥55,515
Vascular complications	¥584,013		¥1,514,380
Atrial fibrillation	¥367,480		¥181,276
Major bleeding	¥652,892	^31	¥652,892	^31
Rehospitalization	¥1,175,859	Claims data analysis^d	¥1,197,655	Claims data analysis^d
Post-operative costs (assumed to be the same for patients who underwent TAVI with SAPIEN 3 and SAVR)^e
	Cost		Source
Disabling stroke	¥405,200		^32
Dialysis	¥400,000		^33
Rehospitalization	¥854,947		Claims data analysis^f

^aBased on the median of SAVR device cost in the MDV analysis (2019–2022Q2).

^bBased on the MDV regression analysis (2019–2022Q2).

^cExcluding device, procedure and complications.

^dBased on the rehospitalization cost within 30 days from discharge in MDV analysis (2019-2022Q2).

^eThis assumption is consistent with previous work.³⁴

^fBased on the hospitalization cost with heart failure in MDV analysis (2019–2022Q2).

Abbreviations. MDV: Medical Data Vision; SAVR: Surgical Aortic Valve Replacement; TAVI: Transcatheter Aortic Valve Implantation.

Implant and procedure costs were sourced from the medical fee information available from the Ministry of Health, Labor and Welfare (MHLW).^29,30 These costs were obtained using revised reimbursement rates and adjusted to 2022.³⁵ The incremental costs of procedural complications were determined from analysis of MDV claims. Specifically, patients who received TAVI or SAVR with a diagnosis of AS between 2019 and 2022 were identified, and incremental expenses associated with complications were extracted and adjusted to the 2022 level. Cost of in-hospital hemorrhage/major bleeding was obtained from the literature³¹ and adjusted similarly. Data on long-term follow-up costs in the Japanese setting related to disabling stroke and dialysis and cost of hemorrhage occurring within hospitalization were not available through the MDV database, so these values were derived using a targeted literature search approach^32,33 and adjusted to the most recent available year (2020) based on the medical care consumer price index tables.³⁶ Rehospitalization cost was obtained from an MDV-based analysis, in which hospitalizations with a heart failure diagnosis were identified, and the average cost for the inpatient stay was determined. Post-operative costs were assumed to be the same for patients who underwent TAVI with SAPIEN 3 and SAVR, consistent with previous work.³⁴ QALYs and costs were discounted at an annual rate of 2.0%, according to C2H guidelines.¹⁶ Specific costs attributable to moderate/severe PVL were not available in MDV or from the literature.

Analytical methods

The cost-effectiveness of TAVI relative to SAVR for intermediate- and low-risk patient populations was determined by the incremental cost-effectiveness ratio (ICER). The ICER threshold for cost-effectiveness was set to ¥5 million/QALY, in accordance with the health technology assessment (HTA) system in Japan.^15,37

A one-way deterministic sensitivity analysis (OWSA) was conducted to investigate the potential impact of uncertain parameters and assess the level of uncertainty in the outcomes of the model. Minimal and maximum values for one-way sensitivity analysis were obtained from the SLR for variables when available, otherwise predetermined upper and lower bounds were calculated for each parameter (Supplemental Table 5). A range of discount rates was also assessed. A complementary probabilistic sensitivity analysis (PSA) to establish robustness of the base case results was conducted using Monte Carlo simulation with 1000 iterations (Supplemental Table 6). The models were run using Excel with Visual Basic for Applications.

Two scenario analyses were conducted: one to test the model outcomes based on a younger average patient age, and the other to test different assumptions regarding the long-term risk of mortality between TAVI and SAVR. A younger patient population scenario analysis was conducted to examine the impact of a younger average patient age of 77.5 years, which is the midpoint of the age range of 75–80 years in which TAVI or SAVR are recommended by recent Japanese Circulation Society guidelines.³⁸ A second scenario analysis was conducted with two different assumptions to explore the possibility that long-term mortality could be different between individuals who underwent TAVI and those who underwent SAVR; it is prudent to explore the impact of different long-term mortality assumptions because only one publication has examined long-term mortality following SAPIEN 3 compared to SAVR.³⁹ Under the first assumption, long-term mortality beyond the trial follow-up was considered to be higher for TAVI than SAVR (hazard ratio [HR] = 1.09), based on PARTNER 2 trial five-year follow-up, which examined the earlier generation SAPIEN XT (Edwards Lifesciences, Irvine, CA) compared to SAVR.⁴⁰ Under the second assumption, long-term mortality beyond two years after treatment was considered to be lower for TAVI than SAVR (HR = 0.9). This was based on the PARTNER 2 S3i trial five-year follow-up, which considered SAPIEN 3 compared to SAVR.⁴¹

Results

Base-case analysis

In the intermediate-risk population, TAVI resulted in a gain of 5.3 lifetime total QALYs, while SAVR resulted in a gain of 4.4 lifetime total QALYs (Table 3). The expected lifetime costs were ¥8.0 million for TAVI and ¥10.2 million for SAVR. As a result, the TAVI was a dominant procedure (TAVI had lower cost and higher effectiveness). In the low-risk population, TAVI led to a gain of lifetime total 9.0 life-years (LYs), while SAVR resulted in a gain of 8.6 lifetime total LYs. The expected lifetime costs were ¥6.7 million for TAVI and ¥6.5 million for SAVR. Hence, the ICER of TAVI compared to SAVR in patients with low risk level was ¥0.75 million/QALY. These ICER values were considerably lower than the predetermined cost-effectiveness threshold of ¥5 million/QALY gained.¹⁵ Therefore, TAVI was found to be economically dominant in the intermediate-risk population and cost-effective in the low-risk population, when compared to SAVR.

Table 3. Base case analysis.

	Lifetime Total Costs (¥)	Lifetime Total QALYs	Lifetime LYs	ICER (¥/QALY) (=ΔCosts/ ΔQALYs)
Intermediate-risk patients
SAVR	10,155,688	4.37	6.41
TAVI	7,973,013	5.33	7.48
Incremental value	−2,182,674	0.96	1.07	Dominant
Low-risk patients
SAVR	6,491,360	6.45	8.62
TAVI	6,710,900	6.74	9.01
Incremental value	219,540	0.29	0.39	750,417

Abbreviations. ICER: Incremental Cost-Effectiveness Ratio; LY: Life Year; QALY: Quality-Adjusted Life Year; SAVR: Surgical Aortic Valve Replacement; TAVI: Transcatheter Aortic Valve Replacement.

Sensitivity analysis

The OWSA allows for the investigation of the sensitivity of our model-based results to variations in specific parameters. The range of each parameter that is tested corresponds to a pre-specified range of possible values for that input. The tornado diagrams visually display that range of the ICER based on higher or lower values of each input, and this is ordered from widest variation (greatest impact on the outcome) to lowest variation in ICER. TAVI remained a dominant procedure in the intermediate-risk population after varying the input values for device cost (TAVI or SAVR), procedure costs, post-operative mortality, long-term cost of disabling stroke, and discount rate (Figure 2A). The intermediate-risk model was most sensitive to the cost of the TAVI implant.

Figure 2. Tornado diagrams of deterministic sensitivity analysis. A. Patients with intermediate surgical risk. B. Patients with low surgical risk. Abbreviations. QALY, Quality-Adjusted Life Year; SAVR, Surgical Aortic Valve Replacement; TAVI, Transcatheter Aortic Valve Implantation.

In the low-risk population, the ICER for TAVI versus SAVR remained cost-effective after varying the input values for device cost (TAVI or SAVR), procedure costs, post-operative mortality, long-term cost of stroke care, and discount rate (Figure 2B). The low-risk model was most sensitive to the cost of the implant, whether TAVI or SAVR. For the full deterministic sensitivity analysis results and breakdown of the costs incurred during hospitalization versus during follow-up, see Supplemental Tables 7–8 and Supplemental Figure 3.

In the probabilistic sensitivity analysis, 100% of simulations for the intermediate-risk population were cost-effective at all willingness-to-pay (WTP) thresholds (Figure 3A). In the low-risk population, 86%, 92% and 96% of the simulations were cost effective at WTP thresholds of ¥2.5, 5 and 10 million/QALY, respectively (Figure 3B). These results indicate that TAVI is highly likely to be cost-effective compared to SAVR in both patient populations.

Figure 3. Probabilistic sensitivity analysis. A. Patients with intermediate surgical risk. B. Patients with low surgical risk. Abbreviations. QALY, Quality-Adjusted Life-Years; WTP, Willingness-To-Pay.

Scenario analyses

In the scenario analysis, in which starting age was adjusted from 83 years to 77.5 years, TAVI was a dominant (cost-saving) procedure in intermediate-risk patients, and cost-effective in low-risk patients (Table 4). In the scenario analysis designed to test the effect of varying long-term mortality benefits, TAVI remained dominant under both mortality risk assumptions for patients at intermediate surgical risk. Furthermore, TAVI remained cost-effective for low-risk patients under both mortality risk assumptions with ICERs of ¥1,989,801 and ¥435,484/QALY for mortality HR of 1.09 and 0.9, respectively (Table 5).

Table 4. Scenario analysis: Testing different patient starting age.

	Starting age: 83 years (Base Case)			Starting age: 77.5 years^a
	Total Costs (¥)	Total QALYs	ICER (¥/QALY) (=ΔCosts/ΔQALYs)	Total Costs (¥)	Total QALYs	ICER (¥/QALY) (=ΔCosts/ΔQALYs)
Intermediate-risk patients
SAVR	10,155,688	4.37		11,545,830	5.96
TAVI	7,973,013	5.33		8,428,640	7.29
Incremental value	−2,182,674	0.96	Dominant	−3,149,321	1.33	Dominant
Low-risk patients
SAVR	6,491,360	6.45		6,630,181	8.09
TAVI	6,710,900	6.74		6,794,613	8.46
Incremental value	219,540	0.29	750,417	164,431	0.38	437,412

^aModeling from 77.5 years to represent the mid-point between 75–80 years, the “grey area” in the Japanese Circulation Society Clinical Guidelines.³⁸

Abbreviations. ICER, Incremental Cost-Effectiveness Ratio; QALY, Quality-Adjusted Life-Year; SAVR, Surgical Aortic Valve Replacement; TAVI, Transcatheter Aortic Valve Implantation.

Table 5. Scenario analysis: Testing different long-term mortality benefits.

	Long-Term Mortality HR = 1 (Base Case)			Long-Term Mortality HR = 1.09^a (Scenario 1)			Long Term Mortality HR = 0.9^b (Scenario 2)
	Total Costs (¥)	Total QALYs	ICER (¥/QALY) (=ΔCosts/ΔQALYs)	Total Costs (¥)	Total QALYs	ICER (¥/QALY)^c (=ΔCosts/ΔQALYs)	Total Costs (¥)	Total QALYs	ICER (¥/QALY) (=ΔCosts/ΔQALYs)
Intermediate-risk patients
SAVR	10,155,688	4.37		10,155,688	4.37		10,155,688	4.37
TAVI	7,973,013	5.33		7,913,529	5.11		8,048,327	5.60
Incremental value	−2,182,674	0.96	Dominant	−2,242,159	0.75	Dominant	−2,107,360	1.20	Dominant
Low-risk patients
SAVR	6,491,360	6.45		6,491,360	6.45		6,491,360	6.45
TAVI	6,710,900	6.74		6,701,967	6.56		6,722,268	6.98
Incremental value	219,540	0.29	750,417	210,608	0.11	1,989,801	230,908	0.53	435,484

^aAssumes long-term mortality hazard for TAVI vs SAVR (through lifetime) is 1.09, based on the 5-year hazard ratio reported in the PARTNER 2 5-year publication.⁴²

^bAssumes long-term mortality hazard is 0.9, based on the 5-year hazard ratio reported in the PARTNER 2 S3i 5-year publication.³⁹

Abbreviations. HR, Hazard ratio; ICER, Incremental cost-effectiveness ratio; QALY, Quality-adjusted life year; SAVR, Surgical aortic valve replacement; TAVI, Transcatheter aortic valve implantation.

Discussion

This is the first study of its kind to examine the cost-effectiveness of TAVI using the most current SAPIEN 3 valve technology in Japanese patients at intermediate and low surgical risk, for whom TAVI was only recently approved. The results of this analysis demonstrated that TAVI was a cost-effective treatment option in intermediate- and low-risk patients with ssAS in Japan. With TAVI, reduced post-operative complications and non-procedure hospitalization costs (likely driven by differences in length of hospital stay) drove QALY gains and cost-savings, which greatly offset the device cost of SAPIEN 3 technology. The cost-effectiveness of TAVI persisted under two challenging scenario analyses, one examining a younger patient population and the second testing two different long-term mortality assumptions.

These CEA results are consistent with studies of low- and intermediate-risk patients conducted in Europe^27,43,44 and North America^45–47 that considered TAVI with SAPIEN 3. These previous studies report a wide range of results, depending on the perspective, patient risk level, and TAVI device generation considered. A recent review of TAVI cost effectiveness analyses found that TAVI was cost-effective in all analyses in low surgical risk patients, and TAVI was cost effective in all evaluations using the latest generation of balloon-expandable TAVI devices.⁴⁸ In this work, the SAPIEN 3 valve was prioritized as it represents a currently used technology and has been used in clinical practice in Japan for the last seven years.⁴⁹ With subsequent device generations and improved operator experience, the outcomes of TAVI will continue to improve, along with the economics.⁵⁰

The PARTNER 3 trial did not capture the planned revascularization strategy, so it was not possible to perform a subgroup analysis on addition of percutaneous coronary intervention (PCI). However, we performed a scenario analysis according to whether PCI or CABG was performed during the index hospitalization for TAVI and SAVR patients, respectively (Supplemental Table 9). Patients who underwent TAVI with PCI or SAVR with CABG had slightly higher lifetime total costs compared to those without PCI or CABG, respectively. This is consistent with findings from previous work.⁵¹

Healthcare systems and perspectives vary between countries, making regionally-specific CEAs important to optimizing clinical practice and particularly important in the rapidly aging society of Japan. There are only two previous CEAs of TAVI versus SAVR in Japan, and the findings regarding the cost-effectiveness of TAVI across these studies are inconsistent.^34,52 Of note, both studies relied on data using older generation SAPIEN valve implants and evaluated the cost-effectiveness of TAVI in higher risk populations. Inoue et al. (2020) concluded that TAVI with SAPIEN-XT was cost-effective versus SAVR and standard of care in high-risk and inoperable patients with an ICER of 1,337,525/QALY and 3,460,810/QALY, respectively.³⁴ Kodera et al. (2017) found that TAVI with the first-generation SAPIEN valve was cost-effective versus medical management in inoperable patients with an ICER of 3,918,808/QALY, but that TAVI with SAPIEN XT was not cost-effective versus SAVR in intermediate-risk patients.⁵²

There are a few reasons why our study may have found that TAVI was an economically dominant procedure in the intermediate risk population compared with SAVR. Firstly, the SAPIEN XT valve is an older, 2^nd generation implant, whereas the SAPIEN 3 valve is the most current implant with a smaller profile (which allows for smaller vascular access) and a sealing skirt (which allows for reduced rates of PVL). Accordingly, it would be expected improved clinical outcomes and reduced cost related to the complications such as bleeding, vascular access issues and PVL when using the SAPIEN 3 implant. Additionally, the present work used MDV claims data specifically from individuals diagnosed with AS, whereas complication costs and follow-up costs for TAVI and SAVR used by Kodera et al. were based on a general cardiovascular population and may not reflect the real-world costs and complications associated with ssAS patients.

The present study also differs in its consideration of the impact of trial-reported postoperative complications on long-term costs based on published costs of care, while the study from Inoue et al. relies on expert opinion to account for a smaller scope of long-term costs. Lastly, Kodera et al. determined costs based on a simple accounting of the device and procedure cost and assumed that complication and follow up costs did not differ for TAVI and SAVR, whereas the present study employed trial data combined with claims data to account for non-procedure costs. Using claims data for costs is considered more robust for a cost analysis because it includes information on insurance payment as well as clinical diagnoses and procedure codes to ascertain medical conditions and treatments.⁵³ Taken together, the different costing methodology and assessment of TAVI using a newer generation device with likely better procedural outcomes could account for the discrepancy in findings between our study and that of prior CEAs.

Our sensitivity analyses indicated that cost-effectiveness of TAVI versus SAVR was somewhat sensitive to the cost of SAVR implant and post-SAVR complications. It is known that these costs may vary widely between hospitals. According to the MDV database, the median overall hospitalization cost of SAVR, excluding the device and procedure fee, was ¥3,347,549 with an interquartile range of ¥1,753,677. This is more substantial than the variation in the overall cost of TAVI, which had a median of ¥1,021,489 with an interquartile range of ¥858,383. This suggests that SAVR hospital economics may vary more widely than TAVI in Japanese hospitals (possibly driven by variation in outcomes). Therefore, cost-effectiveness results should be interpreted with the understanding that an institutions’ outcomes and economics may greatly influence the cost-effectiveness of TAVI versus SAVR at individual facilities.

Limitations

The results of this study should be interpreted in the context of several limitations. First, in the examination of intermediate-risk patients, no randomized controlled trial data that focused on TAVI using SAPIEN 3 vs. SAVR were available. That said, a propensity-matched data set derived from the SAVR arm of the PARTNER 2 A trial and the TAVI arm of the PARTNER S3i registry has been used in other analyses and so event rates were derived from this data.¹²

Second, the model does not account for different costs or outcomes for reintervention between TAVI and SAVR in this patient group. Our decision to exclude reintervention costs in our CEA is consistent with other published models^34,52, as there is little evidence directly comparing long-term rates of reintervention between SAVR and TAVI patients or the cost implications of failing bioprosthetic valves. Furthermore, current data from the United Kingdom Transcatheter Aortic Valve Implantation registry has showed excellent long-term transcatheter aortic valve function, with 91% of patients free of structural valve deterioration between 5 and 10 years post-implantation,⁵⁴ thereby suggesting that the difference between SAVR and TAVI implant durability is likely minimal. Since we would assume that valve reintervention would likely be reflected in mortality, the sensitivity analysis in which the mortality HR of SAVR vs. TAVI was varied should account for any substantial differences in reintervention rates between the two treatment strategies. The model also does not account for institutional variations such as the presence of a disciplined heart team, for two reasons: 1) TAVI is rapidly expanding in Japan and 2) a recent analysis showed that outcomes for patients undergoing TAVI in Japan are consistent across hospitals regardless of their annual procedural volume.⁵⁵

Third, the model assumed no utility impact of the moderate/severe PVL health state because no utility values were available in the literature, especially in Japanese patients. This assumption may need reexamination in future research, as one study found a small, non-significant quality of life decrement for patients with moderate/severe PVL based on the Kansas City Cardiomyopathy Questionnaire score, which cannot be used to determine utility.²⁴

Fourth, some recent cost-effectiveness analyses comparing TAVI and SAVR in low-risk patients have included atrial fibrillation as a health state.^43,56 In the development of our model, we had to be parsimonious in the selection of health states relevant to the treatment and the Japanese healthcare setting. We relied on a targeted literature review of existing models as well as expert input specific to the Japanese treatment perspective in the selection of our health states. Our model accounts for the impact of atrial fibrillation by including the incremental cost of atrial fibrillation as a procedural complication in the first cycle of the model, but not as a long-term health state. The inclusion of an atrial fibrillation state in this model would likely increase the cost difference and QALY difference between SAVR and TAVI, given the higher rates of new onset atrial fibrillation in the SAVR arms of the PARTNER 2S3i and PARTNER 3 trials.^11,12 Therefore, our study – which includes atrial fibrillation as a procedural complication rather than a health state – is a conservative estimate of the cost-effectiveness of TAVI.

Fifth, there are limitations inherent to claims database analyses, such as potential inaccuracies or omissions in the source records. Nevertheless, cost information derived from MDV claims data in this study provides important insights by facilitating meaningful cost-effectiveness comparisons in a large sample of Japanese patients with AS. Similarly, the MDV database did not allow stratification of peri-operative costs according to surgical risk level. However, in Japan SAVR has long been approved for all risk levels, whereas TAVI was approved only in high-risk patients until 2021. It is important to note that patients with higher risk incur higher index hospitalization costs.⁵⁷ Therefore, in the MDV data from 2019–2022 that were used to determine costs, it is reasonable to assume that there was a greater proportion of low-risk/low-cost patients undergoing SAVR than TAVI, potentially resulting in underestimation of the costs of SAVR compared to TAVI. As TAVI becomes used more broadly in low-risk patients, the average cost of TAVI across all patients can be expected to decrease, lessening the cost differential between TAVI and SAVR. Further, the scenario analyses showed that cost-effectiveness of TAVI with SAPIEN 3 remains robust across patient ages and long-term mortality assumptions.

Lastly, there are inherent limitations to cost effectiveness analyses, and assumptions are made based on the best available data; these assumptions may not account for all potential differences across patient populations and healthcare systems.

Conclusions

The present study is the first time that the cost-effectiveness of TAVI with SAPIEN 3 compared to SAVR has been studied in intermediate- and low-risk patients from the Japanese healthcare system perspective. TAVI was dominant in intermediate-risk patients and well below the cost-effectiveness threshold in low-risk patients and these results were stable across a variety of scenario analyses and sensitivity analyses. These results suggest that TAVI can provide meaningful value to Japanese patients at a reasonable incremental cost for patients with low surgical risk and potentially cost-saving in patients with intermediate surgical risk.

Transparency

Declaration of financial/other relationships

JK has no conflict of interest.

SB discloses the following: Abiomed (Consultant; Research Grant); Boston Scientific Corp (Advisory Board Member; Research Grant); Edwards Lifesciences (Speaking Honoraria); Medtronic (Speaking Honoraria; Advisory Board member); Shockwave (Advisory Board Member; Consultant); Zoll Medical (Advisory Board Member; Speaking Honoraria).

KT received lecture fees from Abbott Medical Japan Co., Ltd and Edwards Lifesciences, Medtronic, and Daiichi Sankyo Co., Ltd., and is a Proctor of Edwards Lifesciences and Medtronic.

CT, XJ, and KY disclose employment with Edwards Lifesciences.

Author contributions

JK, SB, and CT participated in study design. CT and XJ conducted the statistical analyses. All authors interpreted the results. JK, KT, and SB wrote the manuscript. All authors reviewed and approved the final version of the manuscript.

Data availability and materials

Due to its proprietary nature, supporting data cannot be made openly available. All summary data relevant to these results are present within the report. Further information about the data and conditions for access are available from Medical Data Vision, Co. (https://en.mdv.co.jp).

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Acknowledgements

The authors thank Emily Farra. and Farah Pathan of Boston Strategic Partners, Inc. for editorial contributions and assistance with manuscript preparation, supported by Edwards Lifesciences.

Additional information

Funding

This research was supported by Edwards Lifesciences.