April 5, 2012

Study shows fewer deaths at hospitals that spend more

John Graves, Ph.D.

Study shows fewer deaths at hospitals that spend more

Research by Vanderbilt’s John Graves, Ph.D., and colleagues at MIT and Cornell University involving a large sample of Medicare patients suggests that higher-spending hospitals are associated with fewer deaths.

“At least for emergency, acute patients in our study, overall mortality was reduced 20 percent to 30 percent in higher-spending hospitals. Doing more in the hospital, including being treated in a teaching or high-technology hospital, has a positive impact on outcomes. We found that right up front,” said Graves, an expert in health economics and policy in Vanderbilt’s Department of Preventive Medicine.

John Graves, Ph.D.

John Graves, Ph.D.

The research, released as a working paper this week by the National Bureau of Economics Research, runs counter to current thinking, which suggests hospitals that spend the most on Medicare patients have no better outcomes and no better patient satisfaction than hospitals that spend less, or even much less.

Some researchers suggest Medicare costs nationwide could be reduced 20 percent to30 percent without harming quality of care.

Graves said this study doesn’t discount the idea there is wasteful spending, but it does provide compelling evidence that some hospitals that spend more on acute or emergent care can have better survival outcomes.

“An inefficient hospital, a high acuity hospital and a technologically advanced hospital all will exhibit high cost structures, and each may or may not be better at saving lives. The challenge is being able to ‘unbundle’ the complex cost-mortality association and pinpoint areas that can be improved upon to lower costs without harming quality,” Graves said.

The idea makes intuitive sense, but teasing out the complexities of cost versus quality has been difficult, and some influential research in the last decade has failed to find benefits to spending more.

In the current study, researchers found a unique way to examine the one-year survival rates among Medicare patients and to compare hospitals to one another in a way that appears to reduce major confounding variables: patient self-selection of hospitals and socioeconomic variability.

To do this, they found a “natural experiment” in which ambulance companies with different transport preferences were effectively randomized to patients in emergencies.

A small study of ambulance privatization in New York City suggested that 9-1-1 ambulance services transport similar patients to hospitals with very different spending practices. While private ambulances tend to transport to hospitals where they may be based, public ambulances are likely to transport to closer, often lower-cost hospitals.

Graves said they took this concept of comparing neighbors who are picked up by different services in an emergency and expanded it to the entire United States.

We spent a lot of the paper showing the patients who are neighbors and went to different hospitals because of the ambulance service that showed up,” Graves said. “It was important that this natural experiment was really a flip of the coin, so we spent a lot of time examining that.”

A review of Medicare records revealed that in acutely ill patients, delivery to a hospital that uses more high-cost services reduces the one-year risk of death as much as 30 percent when compared to neighbors who went to lower-cost facilites.

The study was also able to break down, or “unbundle” some of the reasons for that reduction in risk. Treatment in a teaching hospital reduced risk by 4 percent, while using high technology conferred a 4.7 percent risk reduction. High levels of initial treatment intensity conferred the most protection, reducing risk by 18 percent.

Graves said the paper is important because it shows a creative approach can remove major barriers to more accurate cost-effectiveness research; research that is informing decisions in health care reform.

For hospitals like Vanderbilt that are high-cost and high acuity, it is important to bear in mind that the sickest patients are often the most expensive to treat and also self-select to high technology teaching centers, skewing patient outcomes.

While Graves concedes this approach only applies to the care of acute illness, not to chronic care, he hopes it will encourage more researchers to design ways to tease out the complexities of cost and quality in health care. He emphasizes this research needs to be done before cuts are mode.

“I am not necessarily worried about reform, we just have to be careful how and where we cut,” Graves said.

Other authors of this work include: Joseph Doyle, Ph.D., from the MIT Sloan School of Management, Jonathan Gruber, Ph.D., from the MIT Department of Economics, and Samuel Kleiner, Ph.D., from Cornell University’s Department of Policy Analysis and Management.