Author + information
- Paula D. Strassle, MSPH∗ (, )
- Sameer Arora, MD and
- John P. Vavalle, MD, MHS
- ↵∗Department of Epidemiology, University of North Carolina at Chapel Hill, 170 Roseau Hall, CB #7400, 135 Dauer Drive, Chapel Hill, North Carolina 27599
We read with interest the paper by Kundi et al. (1) that assessed trends in transcatheter aortic valve replacement (TAVR) and surgical aortic valve replacement (SAVR) volume, and the effect of TAVR volume on SAVR mortality. Although this question is an important one, there are concerns regarding how procedure volumes were measured and analyzed.
First, categorizing overall volume from 2011 to 2014 and excluding hospitals that did not perform ≥1 TAVR each year (84% of all TAVR hospitals identified) is troublesome. By summarizing volume over multiple years, heterogeneity exists within the volume groups, especially considering the rapid adoption of TAVR both within and across hospitals. Excluding hospitals that were not TAVR centers in 2011 could have also introduced significant selection bias. Although authors are commended for performing a sensitivity analysis where all hospitals were included, TAVR volume at these newer centers is effectively underestimated, as they had 0 procedures for at least 1 year. Analyzing yearly or quarterly volumes (and allowing volume classification to change over time), would minimize these problems and provide more meaningful results.
Additionally, although TAVR was approved in 2011, TAVR procedure codes were not added by CMS until fiscal year 2012 (effective date October 1, 2011), which prevents appropriate coding of TAVR for most of 2011 (2). This causes 2011 TAVR volumes to be significantly underestimated and potentially introduces bias. A simple solution is to just perform analyses beginning in 2012.
Finally, the authors analyzed Medicare data, which would not include all TAVR or SAVR procedures. In a recent study by our team using NIS (National Inpatient Sample) data (a national, all-payer database), we found that at least 9% of TAVR patients and 32% of SAVR patients were not covered by Medicare (3). Similarly, in an analysis linking 2011 to 2014 TVT (Transcatheter Valve Therapy) registry data to Medicare, around 3,000 TAVRs were excluded because the patient did not have Medicare coverage (4). Therefore, both TAVR and SAVR volumes in Medicare are underestimated (SAVR significantly so), and underestimation is not likely to be consistent across hospitals. This limitation must be reported, and researchers should really consider alternative databases when assessing procedure volume.
Please note: Dr. Vavalle is the site principal investigator for the St. Jude PORTICO trial. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- 2019 American College of Cardiology Foundation
- Kundi H.,
- Strom J.B.,
- Valsdottir L.R.,
- et al.
- Centers for Medicare & Medicaid Services
- Arora S.,
- Strassle P.D.,
- Kolte D.,
- et al.