September 27th, 2011
Carotid Stenting: How Steep the Learning Curve?
In an observational study involving Medicare patients undergoing carotid stenting between 2005 and 2007, Dr. Brahmajee Nallamothu and colleagues showed that low annual operator volume and early experience are associated with increased 30-day mortality. CardioExchange Interventional Cardiology moderators Rick Lange and David Hillis have posed the following questions to Dr. Nallamothu:
RL and DH: Did you find clues about whether the reduced mortality achieved by high volume operators resulted from improvement in their techniques or in their patient selection?
BN: Great question. On a surface level, the more-frequent use of embolic protection devices (EPDs) among experienced operators might be construed as evidence that it was greater technical skill that led to better outcomes. But this could be misleading. Experienced operators could base their patient selection on the likelihood that EPDs will be required or could even turn down cases in which EPDs can’t be used. So we feel that both technical improvements and patient selection are potential explanations for the differences we identified.
RL and DH: Mortality with the most experienced operators in your study (1.4%–1.7%) was substantially higher than was observed in the CREST study (0.7%) and in post-approval studies (0.9%–1.1%). Why?
BN: We believe that our mortality findings largely resulted from a less highly selected patient population than in previous studies. For example, the mean age of patients in our study cohort was 76.2. By comparison, the mean age of patients undergoing carotid stenting in CREST was 68.9. This alone could explain much of the differences you note, particularly with the more-experienced operators.
RL and DH: Is the “cost” of training high-volume operators subjecting patients to low-volume operators initially? Is this acceptable? If so, how is it best done?
BN: Again, great question. There certainly is a “cost” to training new operators to perform complex procedures like carotid stenting. As we note in the article, restricting use to experienced operators early during the dissemination of a new technique could harm long-term access to it. The flip side of this debate is that studies such as CASES-PMS have shown that with good training, new operators can perform the procedure with acceptable outcomes. The significant gap we found could be attributable to the fact that we don’t have formal mechanisms in place to ensure that all new operators go through such rigorous programs. I think we can do better.
RL and DH: In the interest of informed consent, should the physician be required to disclose to the patient how many carotid artery stenting procedures he or she has performed? Do you think physicians would be willing to do this?
BN: As a proceduralist myself, I struggle with this answer. But I think the best perspective on this topic is contained in a commentary by Harlan Krumholz published in JAMA last year. It is hard to read that article and not come away with the feeling that such information should be a part of every informed consent.
RL and DH: Some recommend selectively avoiding less-experienced operators unless they can provide acceptable outcomes data or other convincing evidence of proficiency. Given that 30-day mortality was only 2.5% with the lowest-volume operators in your study, how practical is such a requirement?
BN: This is a major point. A surgeon colleague of mine, Justin Dimick, wrote a wonderful piece for JAMA back in 2004, pointing out that for low-volume procedures with modest complication rates, it is almost impossible to judge operator quality from a statistical standpoint, given the problem of small sample sizes. I think that this could be an issue for carotid stenting. One possible solution is to add rates of stroke, which are much higher and also a measure of proficiency, to outcome assessments. Composite measures that combine risk-adjusted outcomes with volume are also a potential solution.
RL and DH: You found that low-volume operators were less likely than higher-volume operators to use embolic protection devices (EPDs). How would you explain this? Could it be that the importance of EBDs was not appreciated during early experience (procedures done in 2005 vs. 2007)? Some low-volume operators must have used EPDs from the outset. Did they demonstrate such a learning curve?
BN: This is a tough question to answer with certainty. The use of EPDs has grown over the years, especially now that their use is required by CMS for reimbursement. Accounting for the use of EPDs in our mortality models mitigated but did not eliminate the statistical significance of our findings between very-low and high-volume operators. Differences in outcome between patients treated early and late during the experiences of their operators also remained significant even when we restricted our analyses to only those patients who received EPDs. So I think EPDs are an important aspect of this difference, but they cannot entirely explain our findings. As a hypothesis, I wonder if operators who were just beginning to perform carotid stenting might have been trying to keep the procedure as simple as possible by avoiding the use of EPDs.