US Claims Data Point to Better Survival With Impella vs IABP in High-Risk PCI

This latest propensity-adjusted analysis also linked Impella to less MI and cardiogenic shock. RCT results still lie years ahead.

US Claims Data Point to Better Survival With Impella vs IABP in High-Risk PCI

In the latest observational study to address Impella’s role—this time using a large US claims database—patients who received the percutaneous left ventricular assist device (PVAD) during high-risk PCI saw better survival as well as less MI and cardiogenic shock than did patients supported by an intra-aortic balloon pump (IABP).

Moreover, as researchers report in the American Journal of Cardiology, Impella use was not associated with increases in bleeding, stroke, or acute kidney injury.

“There were no prior large-scale contemporary studies evaluating Impella-assisted high-risk PCI compared to IABP support in real-world clinical practice,” Alexandra J. Lansky, MD (Yale School of Medicine, New Haven, CT), lead author of the paper, told TCTMD in an email. “With the increasing complexity of patients undergoing PCI, evaluating the risks and benefits of Impella-supported PCI is relevant to our practice and our patients.”

Their analysis, which covers a 3-year period starting in mid-2016, is the largest such study to date, she said, and its “results extend the findings of the original PROTECT II randomized trial leading up to the approval of Impella” by the US Food and Drug Administration in 2015 for elective and urgent high-risk procedures.

Co-author Jeffrey Moses, MD (NewYork-Presbyterian/Columbia University Irving Medical Center, New York, NY), said that a key advantage of this study is that it looks exclusively at high-risk PCI, not cardiogenic shock or a mixture of the two.

“This is a little cleaner [because] it’s more uniform,” Moses said. With cardiogenic shock, it’s not possible to truly tease out what’s driving outcomes in this way, because “all the key variables that determine prognosis that you would [propensity] match with are not in the database.”

Even with the steps they took in their study, as with any observational analysis, there’s still the potential for nuances that might lead to confounding or selection bias, he noted, but the sensitivity analyses included in this paper and the strict inclusion/exclusion criteria targeting a specific population both provide reassurance.

While the field awaits anticipated RCT findings in both high-risk PCI and cardiogenic shock, these latest data join other retrospective analyses seeking to understand the benefit/risk balance of Impella, several of which have not shown the benefits seen here and, indeed, have pointed to opposite conclusions.

Earlier this fall, a study of Medicare data showed that PVAD use in high-risk PCI more than quadrupled between 2013 and 2019, with no apparent effect on survival at a regional level based on uptake. Several reports on mechanical circulatory support (MCS) in the context of cardiogenic shock, an acute condition notoriously hard to study, also have relied on large databases in recent years but suggest worse outcomes with percutaneous support—spurring debate over how much clarity can be gleaned from these observational analyses.

B. Hadley Wilson, MD (Sanger Heart & Vascular Institute, Atrium Health, Charlotte, NC), vice president of the American College of Cardiology, said the current study stands out for the quality of its data source: the very large, well-known Premier Healthcare Database. Though retrospective and not as definitive as a randomized, controlled trial, the analysis provides interesting insights, he said, calling it “thoughtful” and “transparent.”

“Overall,” Wilson told TCTMD, “I think the report is good and strongly suggests Impella can be used thoughtfully and well in these high-risk PCI patients in a real-world setting.”

More Than 2,100 Patients

Using the Premier Healthcare Database, which captures around one-quarter of US hospital admissions, Lansky and colleagues identified 2,156 patients who underwent nonemergent high-risk PCI with Impella (Abiomed; n = 1,447) or IABP (n = 709) support between April 2016 and June 2019. Their study excluded patients presenting with acute STEMI or cardiogenic shock and those requiring more than one MCS device during index hospitalization.

The PVAD-supported patients tended to be older; had higher prevalence of congestive heart failure, multivessel PCI, and chronic total occlusion; were more likely to be male; and were more apt to have Medicare insurance. The IABP-supported patients, on the other hand, had higher prevalence of obesity, acute respiratory failure, and NSTEMI; were more likely to undergo single-vessel PCI; and more often received glycoprotein IIb/IIIa inhibitors and warfarin.

With Impella, unadjusted in-hospital survival was higher as compared with IABP (95.3% vs 91.0%; P  = 0.0002), with lower rates of subsequent MI (2.5% vs 11.9%; P < 0.0001) and cardiogenic shock (8.3% vs 18.9%; P < 0.0001).

The investigators then performed propensity-score adjustment to account for differences in baseline clinical and procedure characteristics, in-hospital medication use, demographic factors, hospital characteristics, and insurance status. Again, the PVAD was associated with better survival as well as less MI and cardiogenic shock.

Impella vs IABP in High-risk PCI: Propensity-Adjusted Outcomes



95% CI







Cardiogenic Shock



Stroke, bleeding requiring transfusion, and acute kidney injury did not differ by MCS type in either unadjusted or adjusted comparisons.

After adjustment, mean hospital length of stay was 3.4 days with Impella and 4.8 days with IABP. Mean hospital costs were 37% higher with the PVAD ($48,784 vs $35,655 with IABP). Among surviving patients only, costs were 43% higher in the Impella group ($47,541 vs $33,240 with IABP; P < 0.0001 for all comparisons).

Cardiogenic shock stood out as the strongest predictor of in-hospital mortality (OR 7.54; 95% CI 4.97-11.45). Others included MI, stroke, and bleeding requiring transfusion.

Claims Data Pros and Cons

In contrast with Lansky et al’s findings, those of the prior observational reports would suggest “the winds are actually blowing the other direction, more or less, about Impella,” said Wilson. But here, “we find that it looks good [for] about every metric measured.”

And though costs were initially higher with the percutaneous device, he added, the researchers make a good point in their paper that this difference would likely level out over time.

“I use Impella a lot,” Wilson said, “and I think it is an excellent support system in the right situation. One certainly needs to be very mindful about using it appropriately, and that’s where the heart team and shared decision-making approach is paramount, . . . because it is more expensive and there is risk with it potentially. We haven’t answered the question ultimately: is the risk worth the benefit?”

All agreed: only an RCT can fully address this unknown.

Claims data sets have both advantages and disadvantages, Lansky noted. On the plus side, “they are large and represent everyday clinical practice,” she said. Yet their contents are “dependent on physician coding, which is not always complete or accurate (miscoding) and never do capture all the differences in presentation or risk factors (unmeasured confounders) that may influence outcomes.”

Lansky said the potential for unmeasured confounders is greater when treatment choices are being guided by the clinical situation, such as in cardiogenic shock, than it is in a more-homogeneous population like the one studied here: nonemergent patients undergoing high-risk PCI. Their paper also included two sensitivity analyses—one excluding STEMI patients and the other both STEMI and cardiogenic shock—which provided consistent findings.

The study, she concluded, “provides further evidence of the benefits of using Impella during high-risk PCI to stabilize hemodynamics, prevent hemodynamic collapse, enable optimal revascularization, and improve clinical outcomes.”

For Moses, too, the current results are compelling. “Frankly, the intensity of the [survival] benefit is a little surprising,” he acknowledged, yet “there’s certainly no safety signal here.” Moreover, he added, the findings are in line with an analysis of the National Cardiovascular Data Registry, published in May 2022, showing higher in-hospital MACE with IABP compared with other forms of MCS, a group predominated by Impella (adjusted OR 1.59; 95% CI 1.32−1.91).

Taken together, existing data suggest enough “equipoise to justify PROTECT IV,” said Moses. The trial is currently recruiting high-risk PCI patients with complex CAD and reduced LV function. It will be a few years before those results arrive, he added, “so we’ve got to work with what we have.”

Lansky predicted that the PROTECT IV trial “will definitively establish the benefit of Impella for this indication.” Also on the way in high-risk PCI is RECOVER IV, which last month got the green light from the FDA.

Meanwhile, for the shock indication, all eyes are on DanGer: investigators published an update to the statistical plan for that trial in the American Heart Journal this week. Lead investigator Jacob Eifer Møller, MD, PhD, DMSc (Copenhagen University Hospital Rigshospitalet, Denmark), told TCTMD that 315 of a planned 360 have now been enrolled and the target number is expected by the spring of 2023.

On Thursday, Abiomed announced the FDA has “accepted and closed” all of the postapproval studies evaluating Impella in high-risk PCI, cardiogenic shock, and right heart failure.

  • Lansky reports serving on the advisory board of Abiomed.
  • Moses and Wilson report no relevant conflicts of interest.



Pierfrancesco Agostoni

1 year ago
The OR and 95% CI for mortality from the propensity adjusted analysis is above 1, this would suggest more harm of Impella than IABP. Or this is a typo (the OR should be inverted) or, if the value is true, than the whole article should be rewritten. For example for MI and cardiogenic shock the very same values values are below 1, thus going exactly in the other direction than survival. Please clarify!