How rankings are compiled open for debate

Determining which hospital to hold accountable for patient deaths is at the heart of the debate over the relevance of two major sources for health-care outcomes information.

The focus is on a patient, typically very sick, who is transferred from a smaller hospital's emergency department to a large and better equipped hospital, then dies within a 30-day period.

The Centers for Medicare and Medicaid Services (CMS), in its Hospital Compare website, attributes the death to the hospital that received the transferred patient unless the patient was admitted first to the smaller hospital.

A national report commissioned by U.S. News & World Report attributes the same death to the first hospital, regardless of whether the patient was admitted to the first hospital or just passed through its emergency department. The report was done by Research Triangle Institute (RTI).

There is an eye-of-the-beholder component to which consumer health-care rankings carry the most weight in a cottage industry. Other rankings include www.ahrq.gov, www.healthgrades.com, www.drscore.com, and other academic or commercially driven websites.

Medicare and RTI used the same data source, called MEDPAR, but rely on different methodologies to derive their differing rankings.

Medicare said attributing the death to the admitting hospital is fair because all hospitals are treated the same way with a risk-adjusted methodology.

"When rates are risk-adjusted, it means that hospitals that usually take care of sicker patients won't have a worse rate just because their patients were sicker when they arrived at the hospital," Medicare said.

However, for N.C. Baptist Hospital, the distinction between the two methodologies is critical in how officials perceive and promote the quality of its care.

For example, Baptist said 40 percent of 76 patients who died of acute myocardial infarction from July 1, 2007, to June 30, 2010, were transferred from other hospitals.

Those 40 patients are accounted for in the Medicare rankings of Baptist, in which it performed worse than the national average in four of six key categories — death rates for heart attack and pneumonia and readmission rates for heart failure and pneumonia.

Medicare considers those categories important because they "are common among the beneficiary population, are associated with substantial mortality and morbidity, and have considerable variation in outcomes across U.S. hospitals. Mortality within 30 days can be strongly influenced by hospital care and the initial transition to outpatient status."

A study of 5,100 heart patients, released Nov. 8 by the American Heart Association, demonstrates the link between heart attack, heart disease and pneumonia.

The study determined pneumonia, and not a deep incision surgical site infection, is the most common serious infection after heart surgery. The study also found that most infections occur about two weeks after surgery, not one week as physicians previously thought.

Analysts say treating the sickest of the sick is an essential part of the mission of trauma centers such as Baptist, so they understand why Medicare attributes some patient-transfer deaths to those hospitals.

But Dr. John McConnell, chief executive of Wake Forest Baptist Medical Center, and Dr. Russell Howerton, its chief medical officer, said the magazine's survey is more relevant because it excludes patient transfers from death rates.

The Medicare study doesn't "consider the fact that we serve more than 20 counties, many of which lack sufficient access to primary health care," McConnell said in an email. "For issues like the reported CMS mortality rates, we need to address concerns that go well beyond the bounds of our medical center."

RTI says on its website that all patient transfers into the hospital have been excluded from mortality calculations since 2007.

"This was done to help avoid mortality rates that might be inflated by transfers of severely ill patients … to tertiary care hospitals. Research has shown that because of their location, some tertiary care hospitals are particularly vulnerable to 'dumping.'

"This change in methodology means that patients legitimately transferred for appropriate care are lost, but it is more important to ensure that each hospital's mortality numbers are not affected by transfers of very sick patients from hospitals unable to properly care for them."

The Baptist executives say RTI's study produces a more comprehensive, independent analysis of quality and other measures because it features 12 adult specialty categories.

Analysts acknowledge there are socio-economic influences that affect the death and readmission rates for very sick patients, and those factors are beyond a hospital's control.

Those include the patient's existing health condition, family health history, ability to obtain or pay for prescription drugs, and ability to get to and pay for a doctor or specialist visit before and after a hospital stay.

Unlike the CMS study, the magazine's study does not factor in readmission rates, but it does count the hospital's reputation as one-third of its overall ranking.

Some analysts, such as David Meyer, a senior partner for Keystone Planning Group LLC of Durham, say the reputation factor can skew the overall rankings in favor of academic medical centers such as Baptist, Duke University Medical Center and UNC Hospitals. RTI is affiliated with Duke University and the University of North Carolina system.

RTI also combines cardiology and heart surgery as one category rather than listing them separately as Medicare does. The combining of the categories helped Duke earn a ranking of "much better than expected" and Baptist a ranking of "better than expected."

However, Duke and UNC, which also take sicker-than-typical regional patients, performed at the national rate in all six categories in the Medicare report, while Baptist finished worse than the national rate in four categories.

Complicating the comparison of the Medicare and U.S. News & World Report rankings is that the magazine has a commercial agenda with its report. That includes touting marketing opportunities for hospitals with its rankings, including buying the right to use the Best Hospital logo in its advertising and to be listed among the magazine's featured hospitals.

"We focus on these tools and reports primarily as an understanding of how we can continually improve the quality of our care," Howerton said. "These tools are for our quality and not tools for our marketing."

Baptist ran full-page ads last week in the Winston-Salem Journal touting its rankings in the U.S. News & World Report study.

Baptist officials also question CMS using billing data instead of medical records.

Dr. Harlan Krumholz, the cardiologist at Yale School of Medicine who helped come up with the statistical modeling for Hospital Compare, said his model — based on billing data — "produces results that are very similar to what you would get if you used medical record data."

"It's true that the billing data has some limitations — but the simple truth is that when characterizing hospital performance, it is an excellent substitute for the medical record data, which is currently not available on a national basis."

Meyer said he understands RTI's reasoning for excluding transferred patients.

"It seems reasonable to me for there to be some adjustment or factoring for patients who die in the ED of a receiving hospital, who are transfers from another hospital," Meyer said. "We don't want hospitals dumping extremely sick patients on another hospital just to improve their statistics — not that that would happen, but to eliminate the temptation.

"There is a benefit of having two different methodologies, because there may not be one right answer. Two methodologies give everyone two different perspectives.

"That is a good thing for consumers and providers," Meyer said. "Still, patients generally go to a hospital in their insurer network and where their physician is on the medical staff."