Hospital PCI-Related Death Rates Don't Predict Future Performance

Annual hospital percutaneous coronary intervention (PCI)-related death rates were poorly related to hospital performance in later years, a study found.

The observed-to-expected (O/E) death ratio during index years was inversely related with the change in O/E ratios the subsequent year (r=−0.65), indicating regression to the mean, reported Alexander Sandhu, MD, MS, of Stanford University in California, and colleagues in JAMA Cardiology.

Whether hospitals had relatively low (O/E ratio ≤1) or relatively high (O/E ratio >1) mortality relative to expectations, the relationship between index-year and later performance remained inverse (r=−0.45 and −0.60, respectively).

"Little of the variation in risk-adjusted mortality was explained by prior performance," Sandhu and colleagues wrote.

The group also noted that a rise in the O/E death ratio from 1.0 to 2.0 in the index year was linked with a higher O/E death ratio of only 0.15 (95% CI 0.02-0.27) in the subsequent year. The study encompassed 67 New York hospitals with a total of 960 hospital-years using the New York Percutaneous Intervention Reporting System from 1998 to 2016.

These findings back concerns that the current all-cause 30-day PCI death measures mainly show random year-to-year variation instead of real quality of care, noted the researchers.

"Public reporting measures should be carefully evaluated to ensure that they appropriately identify outliers based on a true signal of quality and not random year-to-year noise," they wrote. "The annual hospital PCI-related mortality may not be a reliable factor associated with hospital quality to consider in a practice change or when helping patients select high-quality hospitals."

This paper supports widespread concern in the PCI community that death rates are a poor measure of quality, because patient mix is so variable and out of operators' control, said Ehtisham Mahmud, MD, of the University of California San Diego, who was not involved in the study.

"Often patients who are at high risk for bypass surgery undergo PCI and no amount of risk adjustment can adequately correct for that issue," Mahmud told MedPage Today.

Public reporting for PCI has contributed to risk aversion and possible undertreatment of patients who might benefit from revascularization, emphasized Mahmud. "Rather other measures need to be identified to distinguish individual hospitals [and] operators who have both exceptionally good as well as poor outcomes," Mahmud continued.

Saad Ahmad, MD, of the University of Cincinnati, agreed, telling MedPage Today that public reporting may deter doctors from taking on higher complexity cases.

Yet while PCI mortality may be a poor metric for public reporting, hospitals shouldn't stop tracking it, because it still has potential to show problems internally, said Gregory Dehmer, MD, of the Virginia Tech Carilion School of Medicine in Roanoke in an accompanying editorial.

But he also noted that hospitals should use these data cautiously in evaluating individual operators, particularly those with lower volumes.

"Someone with reported high mortality rates [at] 1 year should not immediately be judged a bad operator, because their results in the following year are likely to be better. In some instances, an operator's annual mortality rate is determined not by their judgment or skill but more on how many times they were on call when the patient dying of a myocardial infarction with cardiogenic shock presented," Dehmer wrote.

All PCI-related death should undergo an unbiased internal review process done by not only physicians, but other members of the catheterization and care teams as well, emphasized Dehmer. Moreover, this process should aim to improve performance, not to identify and punish "bad" operators.

Meanwhile, Dehmer argued, public reporting should be meaningful and easy to understand. One approach could be to report PCI mortality as part of a composite metric instead of in isolation, he continued.

Limitations of the study included that patient severity was based on expected mortality calculated from risk adjustment formula characteristics, however, this method does not take into account unmeasured characteristics that may affect patient severity, the researchers noted. Another limitation was the inability to distinguish between changes in coding and actual changes in severity, they added.