Download this article's Factoid (PDF & PPT for Gold Subscribers)
Fewer than half of deaths occurring within the 30 days after percutaneous coronary intervention (PCI) actually arise from the procedure itself. Moreover, information derived from patients’ charts is far more accurate than what can be obtained from death certificates, according to a new report from the Cleveland Clinic scheduled to be published online May 7, 2013, ahead of print in the Journal of the American College of Cardiology.
Investigators led by Mehdi H. Shishehbor, DO, MPH, of the Cleveland Clinic (Cleveland, OH), looked at all patients who died within 30 days of PCI at their tertiary care center from January 2009 to April 2011. Causes of death were ascertained using detailed chart review and classified according to Academic Research Consortium criteria.
Of 4,078 patients who underwent PCI, 81 (2%) died within 30 days. More than half (58%) of these deaths were considered to be from cardiac causes and the rest from non-cardiac causes, mainly related to infection. However, only 42% of all deaths—approximately three-quarters of cardiac deaths—could be traced to complications of PCI. Cardiac deaths judged unrelated to the procedure were primarily connected to pump dysfunction.
Compared with patients with PCI-related deaths, those who died for other reasons were more likely to present with cardiogenic shock (32% vs. 6%) and cardiac arrest (40% vs. 3%). They also had a higher prevalence of New York Heart Association Class IV heart failure prior to hospitalization (83% vs. 53%) and were more likely to die during the index admission (91% vs. 55%; P < 0.01 for each comparison).
Death certificates, available for 53 patients, showed poor correlation with chart review when used to classify deaths as cardiac vs. noncardiac, with an accuracy of 58%.
A Shift Toward Reporting
In a telephone interview, Dr. Shishehbor told TCTMD that these issues are increasingly relevant as the Centers for Medicare and Medicaid Services has begun advocating for a policy of mandatory public reporting. Moreover, reimbursement is growing ever more closely tied to outcome.
The experience of states that currently require reporting of PCI outcomes—New York, Massachusetts, and Pennsylvania—suggests that “patients who presented [there] with acute coronary syndromes were less likely [than those in nearby regions] to undergo left heart catheterization and percutaneous intervention [even though we] know this the right treatment for those kinds of patients,” Dr. Shishehbor said, citing a 2012 paper published in the Journal of the American Medical Association.
To limit the pitfalls of public reporting, he noted, it is important to strive for accuracy. Death certificates, the most common source for information, “are prone to bias. Everybody knows that,” Dr. Shishehbor stressed. Currently, deaths are categorized as cardiac and noncardiac, “but that does not necessarily reflect performance.”
Reporting Flawed Despite Good Intentions
Government agencies had “good intentions to try to improve quality and give some power to patients to be able to select appropriate physicians and institutions, but obviously these things are having a negative impact,” Dr. Shishehbor concluded.
In an editorial accompanying the paper, Duane S. Pinto, MD, MPH, and Yuri B. Pride, MD, of Beth Israel Deaconess Medical Center (Boston, MA), agree that reporting has not yet achieved its goals.
“Public reporting of PCI outcomes as envisioned by providers, patient advocacy groups, and payors has been implemented with the intent that patients would make educated decisions about where to get their health care and providers would make practice improvements and invest in symptoms of care,” they say, noting that reporting also speeds the adoption of quality improvement measures and improves transparency.
“However, measurement of outcomes, enthusiastic efforts, and admirable goals should not be equated with success,” Drs. Pinto and Pride emphasize. “Implicit in public reporting is that the outcomes measures should be reliable and ultimately useful in improving patient outcomes.”
The findings of the current study, they add, “will not surprise interventionalists but should give pause to those who scrutinize 30-day mortality as the endpoint of choice when determining quality of a hospital or provider.” Instead, the focus should be on the entire system of care, they assert, especially because publicly reported mortality data have “less to do with the operators or hospital systems than the public likely thinks it does.”
Irrespective of Cause, ‘Key Endpoint Is Death’
Robert M. Califf, MD, of the Duke University Medical Center (Durham, NC), said the Cleveland Clinic study offers no new approaches to quantifying PCI-related mortality but still adds incremental knowledge about the multifactorial reasons for such deaths.
“I think most places look at all their deaths in-hospital already,” he said in a telephone interview with TCTMD, predicting that public reporting of 30-day death is on its way “because of reimbursement rules about readmissions. . . . It’s very similar to what surgeons went through maybe 20 years ago, where they started out with in-hospital mortality as the key endpoint [but transitioned to 30 days].”
While it is important to categorize deaths, “so far I haven’t had a dead person have any interest in why they died. The key endpoint is death,” Dr. Califf stressed. “It’s very hard to know exactly why someone died at a given moment. Usually there are multiple things happening all at once.”
The main message, he concluded, “is that we should be measuring a lot of things but shouldn’t rely on any 1 thing, especially major outcomes at the level of something like an individual provider,” which can reflect random variation more than performance.
Cardiac deaths included any death due to proximate cardiac cause (such as MI of fatal arrhythmia), unwitnessed deaths, deaths from unknown causes, and all procedure-related deaths (including those related to concomitant treatment). Noncardiac deaths were those related to respiratory, infectious neurological, GI, and hemato-oncological causes.
PCI-related deaths were those resulting from complications of the procedure such as vascular dissection, aneurysm, perforation, bleeding, renal failure, and definite/probable stent thrombosis.