January 14, 2026
Re: National Coverage Analysis (CAG-00430R2)
Thank you for the opportunity to express our views opposing a National Coverage Determination (NCD) for transcatheter aortic valve replacement (TAVR) in asymptomatic severe aortic stenosis. Our recommendation is based on the scientific evidence and implications described below.
The EARLY TAVR trial does not establish a meaningful patient-centered benefit for asymptomatic Medicare beneficiaries
The EARLY TAVR trial demonstrated statistical superiority for a composite primary endpoint that combined death with disabling stroke into one variable that was part of a composite endpoint that also included stroke and unplanned cardiovascular hospitalization. This justification for coverage is flawed because using a composite measure that combines variables that vary greatly in their importance to patients resulted in a misleading conclusion.1 This is especially true in this case because the very tiny difference in all-cause mortality and disabling stroke for the TAVR patients (11%) compared to surveillance patients (10%) was neither meaningful nor anywhere close to statistically significant, and those two outcomes are the most clinically meaningful endpoints for patients.
Composite endpoints are often used to increase event rates and statistical power, but when component outcomes differ substantially in importance to patients, as was the case here, the composite result obscures the absence of benefit for the most important outcomes. In the EARLY TAVR trial, the composite endpoint was driven largely by hospitalizations rather than survival or major disability, and undergoing TAVR within six months in the surveillance arm was prespecified to count as an “unplanned cardiovascular hospitalization.” As a result, approximately one quarter of patients in the control group who were counted as having an unplanned hospitalization had crossed over to intervention prior to a routine follow-up visit. Although this hospitalization was not planned at the start of the trial, it was not an emerging medical crisis, which is what an unplanned hospitalization suggests. It is particularly problematic because symptom self-reports often triggered crossover, and in an open-label trial of asymptomatic patients, self-reported symptoms may be influenced by patient expectations, anxiety, or dissatisfaction with assignment to surveillance.
For example, one of the interventional cardiologists involved in the study, Dr. Hemal Gada, explained that numerous patients signed up for the study because they wanted TAVR, and were disappointed when they were randomized to surveillance. He expressed his view that the substantial number of asymptomatic patients who almost immediately reported mild symptoms after being randomized to the control group reflected that disappointment, a situation that would not be expected in routine clinical care when patients are reassured that intervention is not yet medically necessary. As a co-author of the published study of EARLY TAVR, he disagreed with the way the results were presented.2 The inclusion of early crossover hospitalizations based on self-reported mild symptoms should not have been included in the unplanned hospitalization endpoint and instead should have been studied independently.
In summary, the reported benefit of early TAVR in this trial was driven primarily by lower scores on a composite outcome that inappropriately included protocol-defined hospitalizations to convert to TAVR, not by improvements in health outcomes such as survival or prevention of a major disability.
Open-label design and adjudication awareness introduce bias in key outcome measures
The EARLY TAVR trial was open-label, and the clinical events committee adjudicating outcomes was aware of treatment assignment. This limitation is particularly relevant because the dominant contributor to the composite endpoint was “unplanned hospitalization” for cardiovascular causes, an outcome that frequently depends on clinical judgment rather than objective criteria and in this trial also was substantially influenced by surveillance arm patients’ self-reported mild symptoms leading to conversion to the treatment group. In an asymptomatic population, where symptom perception, patient anxiety, and provider thresholds for hospitalization may differ by treatment assignment, the potential for bias is substantial. As we noted above, patients in the surveillance arm were aware that they had severe aortic stenosis yet were assigned to defer intervention, and a co-author noted in a later commentary that dissatisfaction with remaining in the control group may have influenced symptom reporting, requests for intervention, and downstream hospitalization decisions.2 These limitations, acknowledged by the study authors, should help persuade CMS that the evidence is not sufficiently robust to support national coverage.
The EARLY TAVR trial medical expertise and study participants are not representative of most Medicare patients or medical care
Participants in the EARLY TAVR trial were predominantly low surgical risk, with a mean Society of Thoracic Surgeons Predicted Risk of Mortality (STS-PROM) score of approximately 1.8%. That means that on average, trial participants were expected to have fewer than 2 deaths per 100 patients within 30 days of TAVR. Participants were also highly selected for anatomy suitable for transfemoral TAVR. These are ethically appropriate decisions for the research study, but it means that the study participants are not representative of the general Medicare population.
In addition, the study results are not generalizable to the Medicare population because all procedures were performed in experienced centers with intensive surveillance, rapid access to intervention, and pre-procedural planning established prior to enrollment. We agree with the study authors that this level of expertise and monitoring cannot be assumed to be replicated in routine clinical practice. In addition, the trial population was overwhelmingly White, and possible racial differences further limit the generalizability of the study results to the Medicare population.
CMS coverage decisions must be based on probable benefits to the heterogeneous Medicare population, including older and frailer patients, those with multiple comorbidities, and care settings without trial-level monitoring or rapid access to intervention. Evidence generated under the exceptional conditions of this study are not generalizable to most Medicare beneficiaries and therefore does not justify broad national coverage. Moreover, Medicare coverage of TAVR for asymptomatic patients is especially questionable given that even under these ideal circumstances there was no benefit in terms of survival or debilitating stroke.
Early crossover and other limitations weaken the case for early intervention
In the clinical surveillance arm of the EARLY TAVR trial, patients did not undergo TAVR solely after spontaneous clinical deterioration. Rather, crossover to intervention occurred based on protocol-defined triggers, including development of self-reported symptoms, changes in left ventricular function, abnormal exercise testing, biomarker thresholds, or investigator judgment under intensive monitoring. This resulted in 26.2% of patients randomized to clinical surveillance undergoing TAVR within 6 months, 47.2% by 1 year, and 71.4% by 2 years. As noted above, hospitalization for TAVR during crossover from the clinical surveillance group to intervention, was classified as an unplanned cardiovascular hospitalization.3
During closely monitored surveillance, concerns and the possible need for intervention generally cause anxiety and lower quality of life, whereas prompt intervention results in lower anxiety and higher quality of life scores. These differences are inevitable and are not an indication that the surveillance itself causes harm. Quality-of-life findings in the trial must therefore be interpreted with caution. Temporary declines in Kansas City Cardiomyopathy Questionnaire (KCCQ) scores observed among some surveillance patients occurred in the period immediately preceding crossover and were followed by rapid improvement after TAVR.3 Moreover, as previously noted, patients who joined the trial hoping for access to TAVR and who were disappointed to be randomly assigned to the surveillance group would likely report the types of changes noted in quality of life during surveillance and after treatment, independent of any true clinical benefit or harm.3 Dr. Hemal Gada pointed out that patients randomized to surveillance were aware that they had severe aortic stenosis yet were assigned to defer intervention, and that situation may have generated dissatisfaction, anxiety, or heightened symptom reporting that would not be expected in routine clinical care when patients are reassured that intervention is not yet medically necessary.2 This trial-specific dynamic limits the implications of the quality-of-life data, weakening their relevance to a decision about broad Medicare coverage of an early, irreversible intervention in asymptomatic patients.
Given the problems with the study design, CMS should be reluctant to make a coverage decision that would encourage Medicare asymptomatic beneficiaries to earlier lifelong device implantation, potential re-interventions, and downstream complications without clear evidence of survival or disability benefit. Earlier implantation of a bioprosthetic valve also raises concerns about valve durability and the cumulative risks and costs of repeat interventions over a beneficiary’s lifetime, outcomes not addressed by the EARLY TAVR trial.
Limited long-term durability data support delaying irreversible valve implantation in asymptomatic patients
All implanted bioprosthetic valves deteriorate over time and may eventually require repeat intervention. Contemporary TAVR devices have thus far been studied for up to 5 years, with limited evidence beyond that timeframe.4,5,6 Clear evidence regarding how long TAVR valves last is especially needed to determine if patients who are asymptomatic should undergo TAVR surgery a year or more before it is necessary.
In EARLY TAVR and similar low-risk trials, patients had a mean age in the mid-70s and relatively low surgical risk, suggesting an expected survival of roughly 8-10 years or longer. In this context, performing TAVR earlier in the disease course increases the likelihood that asymptomatic patients may outlive their first transcatheter valve, raising the risk of structural valve deterioration and the need for repeat valve procedures.
At our request, an analysis was conducted by the nationally respected company DeviceEvents which extracts, consolidates, and provides clear metrics on complex medical device adverse events (MDRs) that have been filed with the FDA. Their analysis identified more than 14,000 adverse event reports of perivalvular or paravalvular leakage for TAVR devices. The reports often do not provide data on when the TAVR procedure took place, but these reports, which increase every year, reinforce our concerns that long-term durability of TAVR valves needs to be carefully studied prior to considering Medicare TAVR coverage for asymptomatic patients.
Delaying TAVR until patients become symptomatic preserves the option of valve replacement when clinical benefit is clear, while reducing cumulative lifetime exposure to device failure, repeat intervention, and associated procedural risks. This consideration is particularly relevant for coverage policy decisions involving irreversible device implantation in asymptomatic patients with substantial remaining life expectancy.
Standards for FDA approval differ from standards for Medicare coverage
FDA approval of medical devices is based on a “reasonable assurance” of safety and effectiveness for patients of unspecified age and health status, whereas CMS is statutorily required to determine whether an intervention is “reasonable and necessary” for the Medicare population. The average Medicare patient is approximately 75 years old, 40% are 75 years and older, and 10% are 85 years and older.7 Older Medicare patients are more vulnerable to poor surgical outcomes than the patients in the EARLY TAVR trial, who were a similar average age but were predominantly low surgical risk, with careful selection and intensive monitoring. This is another reason why the results do not provide evidence that coverage is reasonable and necessary for a Medicare population, since the outcomes that matter most to Medicare beneficiaries are survival, independence, quality of life, and avoidance of major disability. Coverage policy should not move ahead of the evidence, particularly for an irreversible intervention that may expose asymptomatic beneficiaries to substantial out-of-pocket costs and downstream financial burden. In the absence of demonstrated patient-centered benefit, expanding coverage also raises concerns about opportunity costs and the efficient use of limited Medicare resources.
Recommendations
Given the current evidence, CMS should:
- Decline to issue or expand a National Coverage Determination for TAVR in asymptomatic severe aortic stenosis. CMS should consider Coverage with Evidence Development only if future data demonstrate a clear benefit in terms of survival and disability, with useful evidence indicating the devices are safe and effective for more years than patients are likely to need them.
- To meet criteria for an NCD, CMS should require additional trials and longer-term follow-up focused on individual variables that include all-cause mortality, disabling stroke, valve durability, and real-world patient-centered outcomes. Until such evidence is available, expanding national coverage would expose Medicare beneficiaries to substantial risk and cost without proven improvement in outcomes that matter most to them.
References
- Freemantle, N., Calvert, M., Wood, J., Eastaugh, J., & Griffin, C. (2003). Composite outcomes in randomized trials: greater precision but with greater uncertainty?. Jama, 289(19), 2554-2559.
- https://cardiovascularbusiness.com/topics/clinical/structural-heart-disease/tavr/cardiologist-early-tavr-asymptomatic-fda-approval
- Généreux, P., Schwartz, A., Oldemeyer, J. B., Pibarot, P., Cohen, D. J., Blanke, P., … & Leon, M. B. (2025). Transcatheter aortic-valve replacement for asymptomatic severe aortic stenosis. New England Journal of Medicine, 392(3), 217-227.
- FDA CoreValve™ Evolut™ R Transcatheter Aortic Valve Delivery Catheter System: https://www.accessdata.fda.gov/cdrh_docs/pdf13/P130021S058d.pdf
- FDA Edwards SAPIEN 3Transcatheter Heart Valve with the Edwards Commander Delivery
- System: https://www.accessdata.fda.gov/cdrh_docs/pdf14/P140031c.pdf
6. Mack, M. J., Leon, M. B., Thourani, V. H., Makkar, R., Kodali, S. K., Russo, M., … & Smith, C. R. (2019). Transcatheter aortic-valve replacement with a balloon-expandable valve in low-risk patients. New England Journal of Medicine, 380(18), 1695-1705. - KFF Medicare Beneficiaries by Age, 2024: https://www.kff.org/medicare/state-indicator/medicare-beneficiaries-by-age/?currentTimeframe=0&sortModel=%7B%22colId%22:%22Location%22,%22sort%22:%22asc%22%7D