Substitution of Transcatheter for Surgical Aortic Valve Replacement: An Observational Study

Dear Editor,

Transcatheter aortic valve replacement (TAVR) has revolutionized care of severe aortic stenosis (AS).¹ As indications expanded from inoperable to intermediate-risk patients,² procedure volumes have grown considerably and are expected to rise even further in coming years relative to surgical aortic valve replacement (SAVR). Yet, despite data demonstrating increasing TAVR use in the U.S. and worldwide,^3,4 critical information is lacking with respect to TAVR’s effects on national population-based trends in AS management. Specifically, little is known about how much of the rise in TAVR is substituting for SAVR versus expanding aortic valve replacement (AVR) to high-risk or inoperable patients, who otherwise would not be treated.

Because clinical registry data lack information on the population at-risk, which is necessary to examine technology substitution (versus expansion), we analyzed inpatient and outpatient medical claims from a 20% random sample of Medicare fee-for-service beneficiaries. We used a diagnosis code-based algorithm (available upon request) to identify beneficiaries with AS. For inclusion, we required that beneficiaries had ≥2 claims for a diagnosis of AS between January 1, 2009 and December 31, 2015. We also required beneficiaries to be ≥66 years of age at the time of their first AS claim, and be continuously enrolled in Parts A/B 1 year before and after this claim (or, for those who died, during the time that they were alive). We excluded beneficiaries with end stage renal disease.

We calculated annual rates of AVR, SAVR, and TAVR. The numerator was the number of unique beneficiaries who underwent a given procedure in a study year. The denominator was the number of beneficiaries with AS that same year. We adjusted these rates for differences in beneficiary age, gender, race/ethnicity, Medicare Accountable Care Organization alignment, and comorbid status (based on hierarchical condition categories) using the observed-to-expected adjustment method. We fit linear regression models to evaluate for temporal changes. All tests were two-tailed, with the probability of Type 1 error set at 0.05. Our study was deemed exempt from Institutional Review Board oversight.

Of 547,589 beneficiaries with AS, 45,181 underwent AVR. The Figure demonstrates a numerical, but non-statistically significant rise in AVR rates (per 100,000 beneficiaries with AS) from 4,044 [95% confidence interval (CI), 3,925–4,163] in 2009 to 4,851 (95% CI, 4,730–4,972) in 2015 (P=0.053 for trend). Between 2011 (Food and Drug Administration approval) and 2015, TAVR rates increased over 20-fold [from 82 (95% CI, 64–101) to 1,863 (95% CI, 1,783–1,943); P=0.001], while SAVR rates decreased [from 4,044 (95% CI, 3,925–4,163) to 2,988 (95% CI, 2,894–3,081); P<0.001]. TAVR accounted for 2.05% of AVRs in 2011 and 38.38% in 2015.

Figure.

Rates of overall AVR, TAVR, and SAVR (per 100,000 beneficiaries with AS).

Our study has limitations that merit discussion. First, we identified the denominator for our rate calculations through provider-coded diagnoses on medical claims. Consequently, errors or changes in coding practices could affect our estimates. However, our focus on AS patients as opposed to all beneficiaries is an advance over prior work. Second, medical claims lack granular clinical information (e.g., echocardiographic data on AS severity), which is necessary to understand how observed trends relate to AS severity and patient selection for TAVR or SAVR.

In conclusion, our study shows a dramatic increase in TAVR use offset by a mild decline in SAVR use for AS such that overall AVR rates have only modestly increased. This is important because TAVR was originally established as an alternate treatment option for inoperable or high-risk SAVR patients. However, our findings suggest that TAVR may have initially replaced SAVR as opposed to increasing the total number of patients undergoing AVR. Since the latter was a key motivation for TAVR’s development,⁵ these trends require monitoring as the technology extends to even lower risk patients.