Procedural Deaths Fall Short as Quality Metric for Hospital Acute MI Care

Nearly half of hospitals judged to be high- or low-performing on procedural metrics were reclassified by a disease-based metric.

Procedural Deaths Fall Short as Quality Metric for Hospital Acute MI Care

Procedural-based outcomes may not provide the most accurate picture of the quality of hospital care for acute MI and may underestimate mortality in the highest-risk patients, new research suggests.

“As cardiologists broadly, and as interventional cardiologists specifically, we're being judged by an increasing array of metrics, and by an increasing array of entities,” said Jay Giri, MD (Hospital of the University of Pennsylvania, Philadelphia), the study’s senior author. Metrics may be publicly reported, like PCI outcomes, or accessed by payers, leading to what Giri calls “layered consequences” that include perceptions about quality of care that may not be accurate, and procedural risk avoidance.

“A common refrain that comes from physicians who are mistrustful of the metrics is that you can put hundreds of checkboxes on a paper about patient characteristics, but you can't necessarily capture who they are completely or what their risk of a given outcome is because some patients can't be accurately risk adjusted,” he added. “I think things can go a little off kilter when the metrics are being used by a variety of entities for different purposes.”

Instead, Giri and colleagues argue in a paper published online this week in JAMA Cardiology, switching to a disease-based mortality metric may provide a better means of assessing overall quality of care. Their findings suggest that the NCDR Chest Pain–MI (disease-based metric) and CathPCI (procedural-based metric) are not that well correlated and that one of the reasons for this lack of correlation is risk avoidance in the highest-risk groups.

“We would hope that people would take a step back and reexamine whether this PCI-related metric is really a good idea,” he said. “The fact that we have an infrastructure in place for this MI/disease-related metric, it seems like at the very minimum shifting toward that should be a focus for public reporting in places where that is being considered or is actively being done.”

Hospital Reclassifications, Mortality Gaps Seen

For the study, led by Ashwin S. Nathan, MD (Hospital of the University of Pennsylvania, Philadelphia), the researchers looked at excess mortality ratios of acute MI patients between 2011 through 2017 at 625 hospitals participating in the NCDR Chest Pain–MI and CathPCI registries. The analysis sought to understand the hospital-level correlation between disease-specific mortality and PCI procedural mortality, and to assess whether switching to a disease-based metric affected hospital performance assessment.

I think things can go a little off kilter when the metrics are being used by a variety of entities for different purposes. Jay Giri

When the hospital data from hospitals in both registries were linked, the correlation between disease-based outcomes and procedural-based outcomes was found to be moderate (P = 0.53).

More than half of all sites (56.7%) in the highest-performing tertile for disease-based risk-adjusted mortality were also in the highest tertile for procedural risk-adjusted mortality. The other 43.3% of sites that had been in the highest-performing tertile were reclassified into a lower category, including 25 hospitals (12%) that ended up in the lowest-performing tertile for procedural risk-adjusted mortality.

On the other hand, 55.8% of hospitals in the lowest-performing tertile for disease-based risk-adjusted mortality were also in the lowest tertile for procedural risk-adjusted mortality. The other 44.2% were bumped into a higher category for procedural risk-adjusted mortality, with 24 sites (11.5%) reclassified into the highest-performing tertile for procedural risk-adjusted mortality.

For the entire cohort, procedural mortality was significantly higher than disease-based mortality, which makes sense, Giri said, since medical therapy and NSTEMI patients not getting procedures would not be counted in the CathPCI database, nor would those who may have eligible for PCI but had it withheld for risk-avoidance reasons. When the researchers sought to examine if the difference between the two databases would go away if only STEMI patients were analyzed, the gap narrowed, but higher mortality with the procedural metric remained.

“Then we had the kicker, where we looked only at patients who had cardiogenic shock or cardiac arrest,” Giri observed. “What we actually saw was very interesting, in that the relationship between the two metrics for mortality flipped.” The difference was statistically significant with lower procedural mortality compared with disease-based mortality in these two high-risk groups (P < 0.001).

“The implication of that to us was that some patients with cardiogenic shock and arrest are being selected to have PCI withheld and then they’re dying,” he added. While some of the withholding is likely to be appropriate, Giri said it does create concern that physicians are being judged by, and responding to, inappropriate metrics.

Time for a More Holistic Approach?

As a whole, the analysis adds more evidence that procedural metrics do not accurately tell the story of all the pieces that go into an individual hospital’s attempts to optimize outcomes for patients with acute MI, Giri said.

“When you're just boiling it down to the procedure, you're ignoring all of those other parts of things. When a patient has a complication as a result of one of those other things or something that couldn’t have been avoided by any of these things, who gets dinged for it? Only the proceduralist,” he observed. “It has consequences of risk avoidance, which is bad for patients and public health.”

According to Giri, moving to a disease-based metric has the potential to incentivize all stakeholders involved in the care within that hospital to focus on quality holistically.

“This can flip everything on its head,” he concluded. “If you have a disease-based metric with a patient dying in front of you, and your only chance to save them is to try to do a PCI or maybe put in some support, now you’re incentivized to be more aggressive, as long as you're doing the right thing [and] incentivize the right overall care patterns.”

Disclosures
  • Nathan reports no relevant conflicts of interest.
  • Giri reports serving on an advisory board for and receiving personal fees from AstraZeneca; grants from ReCor Medical and St Jude Medical; and personal fees from Partners Healthcare outside the submitted work.

Comments