Volume Still an Important Surrogate for TAVR Quality, New Data Suggest

Newly published data from the Society of Thoracic Surgeons/American College of Cardiology Transcatheter Valve Therapy (TVT) Registry support the need to continue looking at volume requirements as the field explores more direct ways to assess the quality of TAVR programs.

The new numbers come 1 week after the Centers for Medicare & Medicaid Services (CMS) released its proposed update to requirements for hospitals wanting to be reimbursed for performing TAVR. The proposal tweaked some criteria, but maintained procedural volume norms for both new and existing programs.

Now, looking at more than 100,000 TAVRs performed from 2015 to 2017, researchers led by Sreekanth Vemulapalli, MD (Duke University Medical Center, Durham, NC), found that procedural volume at the hospital level—both for transfemoral and nontransfemoral TAVR—is inversely associated with 30-day risk-adjusted mortality, even after excluding a hospital’s first year of cases to account for a potential learning curve.

How to juggle concerns over volume and outcomes while improving access to care has been a topic of ongoing debate within the TAVR community. The current study findings contrast with a recent study confined to cases performed with the Sapien 3 valve (Edwards Lifesciences), but are consistent with prior analyses, including one examining earlier data from the TVT Registry.

This new analysis—published online April 3, 2019, ahead of print in the New England Journal of Medicine—is by far the largest to examine the issue and focuses on contemporary practice with the latest valves and techniques, overcoming some of the critiques of earlier work, Vemulapalli told TCTMD.

“Even in this day and age, with all the tremendous improvements in TAVR device technology and technique, we still do see a relationship between volume and outcome,” he said, adding that “it doesn’t appear that this relationship can be explained by a learning curve.”

Vemulapalli said this should be considered by CMS as it comes to a final decision on its updated national coverage determination (NCD) for TAVR, which is due by June 24. Of note, the NEJM elected to publish the results by an “accelerated schedule,” according to the email sent to members of the media.

Study co-author Michael Mack, MD (Baylor Scott and White Heart Hospital, Plano, TX), told TCTMD that even though more direct assessments of TAVR quality would be preferred—and are, in fact, in the works—there is still a pressing need to consider procedural volume in assessing new and existing programs.

“In general, if you measure outcomes of a procedure, the more complex the procedure, the more the volume-outcome relationship matters,” Mack said, adding that questions have appropriately been raised as to whether TAVR has gotten to the point where procedural volume doesn’t have a substantial impact on outcomes. “What this study shows is it’s not there yet. It still is the more you do, the better you are at it.”

Addressing the Controversy

When TAVR was first commercialized, CMS included procedural volume requirements in its 2012 NCD. However, those requirements have been called into question in recent years as technology and techniques have improved and the patient pool has widened to include lower-risk patients.

This new analysis was meant to update a prior study covering TAVRs performed between 2011 and 2015 included in the TVT Registry. Vemulapalli, Mack, and colleagues looked at data on 113,662 TAVRs performed at 55 hospitals by 2,960 operators between 2015 and 2017. Most of the procedures (84.7%) were done with transfemoral access.

Median annualized volume of transfemoral TAVRs at the hospital level was 54. When divided into quartiles, the lowest-volume centers performed an average of 27 a year and the highest-volume centers performed an average of 143.

As a continuous variable, procedural volume for transfemoral TAVR was inversely associated with risk-adjusted mortality at 30 days. The mortality rate was 3.19% in the lowest quartile and 2.66% in the highest (OR 1.21; 95% CI 1.03-1.41). The findings were similar in a sensitivity analysis excluding the first year of transfemoral TAVRs at each center.

Although annualized procedural volume was not associated with risk of overall complications at 30 days, there was a higher risk of a composite of major vascular complications or major bleeding among centers performing the fewest procedures versus those performing the most (10.03% vs 8.21%; OR 1.25; 95% CI 1.08-1.45).

The investigators also evaluated operator volume and nontransfemoral TAVR volume, finding that both were inversely associated with 30-day risk-adjusted mortality.

How Best to Assess Quality

Asked whether these findings support the changes that CMS has proposed—which include relaxing some volume requirements, especially for hospitals looking to start a TAVR program—Vemulapalli responded, “If the idea is to use volume as a marker of quality, which is what I think CMS is trying to do here, then maybe lowering [volume thresholds] isn’t such a good idea based on what we saw, which is that there is this correlation and it looks like it’s at least 50 cases at an operator level before the effect of volume starts to flatten out. So lowering the values might not be the most consistent thing with our findings.”

He acknowledged that CMS is trying to find the right balance between ensuring quality of care and maintaining sufficient access to TAVR, but said that the “reasonable, theoretical concerns” about an adverse impact on access from maintaining volume requirements are not supported by existing data.

“It’s not immediately clear whether lowering these thresholds would actually improve access or not. It’s a reasonable hypothesis that it would. We just don’t know. We don’t have enough data to know if there is a group of patients out there right now who’s not getting TAVR that if we lowered those thresholds they would then be able to get TAVR,” Vemulapalli said.

Mack said it’s appropriate to keep some volume requirements in place right now, at least until a strong enough metric to more directly assess quality can be developed. Volume, he added, “is not a perfect surrogate [for quality], but it’s the best that we have.”

Consistent with what many others have expressed throughout this debate, Mack said, “We would love to get beyond volume and get to a pure quality metric, and indeed the TVT Registry has developed a quality metric for prediction of mortality at a year. So where all this is hopefully going over the next year-and-a-half is to go to a quality metric that’s a three-star rating just like STS surgery outcomes and then public reporting of the outcomes.”

Mack and Vemulapalli both stressed that even when moving to a direct measure of quality, a certain procedural volume is required to make assessments valid.

“The lower the numbers that a hospital is doing, the harder it is to estimate how good they’re actually doing because there’s more of a play of chance,” Vemulapalli explained. Indeed, in the current analysis, mortality rates were more variable in low-volume centers.

“Mandatory public reporting of actual outcomes, with a lower threshold limit of volume, would be a better way or another potential way to actually assess quality,” he said, “and it would remove some of the issue of these limits on how much volume you need and that might help to answer the access question.”