A 30-Day Mortality Rate Mystery

Daily death rates for cardiac surgery patients in one large study jumped at the 30-day mark. Was it because physicians were treating the performance measure instead of the patient? What are the unintended consequences of quality of care and performance measures, anyway?

Cheryl Clark is on vacation.

From time to time, Bryan Maxwell, MD, hears colleagues jokingly remark that a decision about a patient's care will have to wait 30 days, as in the 30-day mortality rate, a standard quality of care measure.

Maxwell, an anesthesiologist and critical care specialist at Johns Hopkins, says he has begun to wonder whether these doctors are only half joking. Are they making care decisions based on the needs of the patients? Or is clinical care being guided by the push to meet performance measures?

"I became interested in looking at it because in anecdotal, informal, unofficial, conversations with people, that idea would come up that way," says Maxwell. In a study published in Health Services Research, he and his team found daily death rates for about 600,000 cardiac surgery patients jumped at the 30-day mark.

Maxwell, then at Stanford University and now at Johns Hopkins, concluded that "the abrupt increase in mortality at day 30 has no clear organic etiology. This analysis raises the possibility that this observed shift may be related to clinician behavior because of the use of 30-day mortality as a quality metric, but further studies will be required to establish causality."

No one disagrees on the need for more research. But some bristle at the notion that the 30-day mortality rate could be a death panel in reverse.

What doctor would take measures to keep a dying patient alive for a few extra weeks to avoid a "worse than US national rate" scold on the Hospital Compare website? Still, regardless of what drove the timely spike in cardiac surgery deaths, the study raises the issue: What are the unintended consequences of quality of care and performance measures?

Actions 'Not in the Patients' Best Interest'
A qualitative study from the University of Minnesota attempted to answer that question by looking at performance measures for primary care at Veterans Administration hospitals. Using at data from 2009, the researchers, led by Katie White, EdD, MBA, looked at screening measures for obesity, hypertension, cancer, vascular disease, diabetes, mental health and substance use.

They interviewed primary care staff and facility leaders to help "identify unintended consequences and explore the pathways by which they arise." The research, published last year, uncovered a type of clinical care providers described as "'treating the measure, not the patient.'

A key finding: "The perception that performance scores must be improved or that a provider or clinic is not in compliance with expected targets can lead providers to take actions when they are not in the patient's best interest." Examples include overuse of medications to improve blood pressure. In some cases, the time it takes to deal with performance measure prevents providers from addressing "unmeasured, but important aspects of patient care," the researchers wrote.

Coming up short on a consumer ranking site like Hospital Compare is one thing. But, with the shift toward value-based purchasing, the stakes are getting much higher. In a November 2013 paper in the New England Journal of Medicine, Robert Berenson, MD, of the Urban Institute warned that Medicare's pay-for-performance program for doctors is a flawed approach that "cannot accurately measure any physician's overall value" and therefore won't hold the physician accountable for quality.

Without data on the efficacy of these measures— or how physicians will respond to them—"policy makers are prematurely deciding that this is a correct approach," Berenson wrote. He had little doubt that the 30-day jump in mortality observed by Maxwell's team is related to performance measures: "There's no other explanation." And hospitals, he said, need to take note.

It's 'Hard to Game the Date of Death'
Focusing on one outcome, such as readmissions, to avoid a specific penalty may compromise quality of care, safety, and patient satisfaction for other clinical problems, according to a March opinion piece in Journal of the American Medical Association.

But, Maxwell's paper does not make the case that this is driving the mortality spike he observed at 30-days, says David Goodman, MD, one of the three researchers from the Dartmouth Institute for Health Policy who authored the piece: "It is hard to game the date of death."

At the same time, Goodman and his co-authors agree that performance measures need to be refined. In the JAMA paper, they argue that Medicare should encourage hospitals to "invest in broader goals and should reward success."

For example, they suggest that readmissions reduction programs should consider admissions as well because many of the behaviors that lead to unnecessary admissions also lead to readmissions. They cite a program that targeted care transitions in 14 communities in an effort to reduce readmissions. The per capita rate of both admissions and readmissions declined when the hospital worked with community providers to improve the coordination of primary care.

"This unexpected result—a positive variation on the law of unintended consequences—teaches an important lesson: targeting the performance of one medical care service can change a behavior that can affect the performance of other services," they wrote.

They key, Goodman says, is to develop performance measures that reflect what's most important for patients and their care.

"Performance measures are sometimes a compromise between what is important and what can be measured," he said. "The need to pay for quality and outcomes is important. It is hard work to get the measures and the incentives right. We're not there yet. But we need to continue on that path."