TUESDAY, March 21, 2006 (HealthDay News) -- A new statistical method that uses ordinary billing records to assess a hospital's quality of care may boost the treatment of heart attack and heart failure, researchers report.
Differences between hospitals in terms of the care patients receive do make a big difference. The study found that post-heart attack death was significantly more likely at low- vs. high-rated centers.
So far, however, there have been a couple of major problems in assessing how well individual hospitals do in treating those conditions, noted Dr. Harlan M. Krumholz, professor of medicine at Yale University and lead author of two reports on the new method in the March 21 issue of Circulation.
One problem is that results depend on how sick the patients were to begin with. The other is that complete medical records about those patients aren't always available.
"How likely people are to survive -- we know it isn't wholly dependent on the hospital," Krumholz said. "But at least some of it is due to the quality of care provided. We needed a way to see how sick patients were so we could level the playing field. This was a challenge because the quality of data we have about hospitals in this country is not ideal."
For heart attacks, Krumholz's team examined medical records on more than 180,000 Medicare patients gathered during 1994 and 1995 for a separate study. The researchers compared that information with Medicare billing records to identify 27 factors related to the risk of death from heart attack and 24 related to the risk of death for heart failure -- factors such as age, severity of illness and presence of other conditions such as diabetes.
"We employed the best statistical approaches in this effort and have produced models that can characterize the impact of hospital quality on the patients' likelihood of survival," Sharon-Lise T. Normand, professor of health-care policy at Harvard Medical School, and a member of the research team, said in a prepared statement.
Adjusting for patient differences, the researchers found that heart-attack patients admitted to the top 5 percent of 4,500 U.S. hospitals had a 15 percent risk of dying in the next 30 days. The 30-day death rate for hospitals in the bottom 5 percent was 20 percent.
For heart failure, the 30-day death rate was 10 percent for the top hospitals, 14 percent for those at the bottom 5 percent.
The bottom line: "For every 20 heart attack patients admitted to a bottom-rated hospital rather than to a top-rated hospital, there is one additional death," Krumholz said.
The job now is to put the information gathered in the study to practical use, Krumholz said. There are commercial programs that evaluate hospital performance, but their methods are proprietary and not open to scrutiny, he said.
"Our approach is to be transparent," Krumholz said. "We are going to allow hospitals to look at mortality data to see how they compare to other hospitals. We will look hard at hospitals who are excelling, and that will give us a chance to learn what the differences are between the hospitals that are doing well and those that are not doing well. By making these differences visible, we will add motivation to doctors and hospitals."
The new methodology has been endorsed by the National Quality Forum, a not-for-profit group whose purpose is to improve medical care in this country, Krumholz said.
The studies were done in collaboration with Medicare and the Centers for Medicare and Medicaid Services, he added, which "are considering whether to use this to publicly report hospital performance on the Web."
The study will be a great help to hospitals in assessing the quality of their care, said Dr. Clyde Yancey, professor of medicine at the University of Texas Southwestern Medical Center at Dallas, and a spokesman for the American Heart Association.
"Every hospital has its strategies to improve care, but it is difficult to prove that the best application of those strategies improves outcome," Yancey said. "This really smoothes every hospital in the country into one model. They can see whether their strategies are working. This gives hospitals a chance to know what may be missing in their quality investigations."
The big surprise of the studies is that information of use in medical practice could emerge from non-clinical, administrative data that is much more accessible than clinical data, Yancey said. "It is provocative that a risk-predictive model using administrative data should be so closely correlated to clinical results," he said.
More information
The American Heart Association has more about the signs and symptoms of a heart attack.
