CardioExchange Archive

CardioExchange welcomes Dr. Steven Marso to answer questions about his recent paper in JAMA, which analyzed data from 1.5 million PCI patients from hospitals enrolled in the National Cardiovascular Data Registry (NCDR) to examine the use of bleeding-avoidance strategies. Manual compression was used in 35% of cases, vascular closure devices (VCDs) in 24%, bivalirudin in 23%, and the combination of a VCD and bivalirudin in 18%. The rate of bleeding was 2.8% for manual compression, 2.1% for closure devices, 1.6% for bivalirudin, and 0.9% for combination therapy. Marso et al reported an unusual paradox in which patients at high risk for bleeding were less likely to receive combination therapy. They conclude that their findings “emphasize the opportunity to improve the safety of PCI and to further explore cost efficacy by directing such strategies to those patients most likely to benefit from them.”

A large number of catheterization centers were included in this registry and, in turn, a very impressive number of patients were included in this study. To get a better sense of where these data come from and what is captured, are there certain types of centers (e.g., large academic versus small community) that tend to participate in this registry and certain types that tend not to? 

This analysis was of 1.5 million individuals treated at 955 U.S. hospitals. We believe this was a fair representation of U.S. PCI centers, however, it’s very difficult to know whether or not this is true. About 39% of the centers are from the South, 33% from the Midwest, 12% from the Northeast, and 16% from the West. Given the population distribution, you may guess that the Northeast may be somewhat under-represented in this analysis. Nine percent were university centers and approximately 89% were private/community centers. A very good question is, “Who is not represented in the ACC NCDR CathPCI Registry?” It would be fascinating to look at all the PCI centers in the U.S. to determine whether or not the registry is reasonably representative, and I suspect we would find differences. To my knowledge, this work has not yet been done.

Given the regional differences in use of VCDs, do you think that lower use of these devices in some regions is driven by cost concerns? Do you think this could be an opportunity for region-specific, cost-benefit analyses?

I think the variability in VCD use is driven by a number of factors. First, VCDs were evaluated in previous work to facilitate patient ambulation. Operators may value this outcome measure differently; thus, use very likely varies across U.S. PCI centers. Ambulation time is certainly important to some patients for whom down time is uncomfortable. Second, centers may also utilize VCDs in order to optimize patient flow. Third, cost considerations are also important. Although variable, VCDs range in price from $150.00-250.00, depending on regional contracts. If you have a very high-volume PCI center, this cost is not insignificant. The challenge is to apply VCDs in patients while also minimizing cost. I believe our study has shown a potential way to achieve this goal. For example, if VCDs and bivalirudin are used in patients at intermediate or high risk for bleeding, there might be cost savings or, at the very least, cost neutrality in selected populations.

Relative renal dysfunction is repeatedly reported as one of the strongest risk factors for periprocedural bleeding. Did any of your analyses consider patients with mild or moderate renal dysfunction as opposed to those with frank renal failure?

We utilized the NCDR CathPCI bleeding risk model, which consists of 9 clinical variables, one of which is renal function as measured by estimated glomerular filtration rate (modification of diet in renal disease method). This approach to bleeding-risk stratification incorporates the entire continuum of mild, moderate, and severe renal dysfunction.

This is a great example of how registry data can shed light where trials cannot or have not. Based on the quality of these data, do you think that practice guidelines should change to take into account these results? If not yet, what further work do you think needs to be done?

That decision, of course, lies with the physicians and other professionals who draft national guidelines and recommendations. I do believe registry analyses like these provide relevant clinical information, for which data from randomized trials is lacking. There are two aspects of this study that I think deserve mention: First, we risk-stratified individual patients in a way that allows physicians to provide individualized or tailored therapy based upon this risk. Thus, therapies can be directed to those at greatest risk and, therefore, most likely to benefit. Second, we evaluated two therapies not likely to be studied again in clinical trials. VCDs and bivalirudin have been individually studied and it would be expensive to repeat these studies in such a broad population and in a head-to-head comparison. Therefore, I think this set of data is unique in that it provides insight into PCI management that will be difficult to obtain from randomized, controlled trials.