The same scientists found the drug also raised the risk of kidney failure, heart attacks and strokes in a study published last year. Most of the deaths in the new study were related to those problems.

    The researchers said replacing the drug with other, cheaper medications for a year would prevent 10,000 deaths worldwide over the next five years.

    The study, published in Wednesday's Journal of the American Medical Association, intensifies questions about how best to track the safety of drugs after they have gone on the market. The Food and Drug Administration approved aprotinin in 1993.

    Last year's study and other research led the FDA to review Trasylol's safety and order stronger warning labels in December. But that action wasn't strong enough, said Dr. Dennis Mangano of the nonprofit Ischemia Research and Education Foundation, lead author of both studies.

    "I believe that for the vast majority of coronary bypass patients the drug should not be used," Mangano said. But he said the drug should remain on the market because some very high-risk patients may benefit from it.

    Mangano said patients should ask their doctors what drug, if any, they will be given to slow bleeding during heart surgery and about the risks. Past patients should find out if they have been given aprotinin so their doctors can watch for problems, he said.

    Bayer said in a statement the findings are unreliable because Trasylol tends to be used in more complex operations and the researchers' statistical analysis did not fully account for the complexity of the surgery cases.

    The drug company said it will "work with regulatory agencies and external experts in the field to further evaluate the findings."

    The FDA is already reviewing aprotinin's safety. The new study is an important contribution to that review, which could result in additional warnings, said Dr. Dwaine Rieves, deputy director in the FDA's Division of Medical Imaging and Hematology Products.

    The study followed 3,876 patients who had heart bypass surgery at 62 medical centers in 16 nations. Researchers compared patients who received aprotinin to patients who got other drugs or no anti-bleeding drugs. Over five years, 20.8 percent of the aprotinin patients died, versus 12.7 percent of the patients who received no anti-bleeding drug.

    When researchers adjusted for other factors, they found that patients who got Trasylol ran a 48 percent higher risk of dying in the five years afterward.

    The other drugs, both cheaper generics, did not raise the risk of death significantly.

    Dr. Brett Sheridan, a heart surgeon with the University of North Carolina Health Care System who was not involved in the study, said several years ago he quit giving aprotinin almost entirely because he had seen more kidney damage in patients who got the drug.

    "It's not a perfect study, but the data are compelling enough that we have to use aprotinin judiciously," Sheridan said of the new research.

    The study was not a randomized trial, meaning that it did not randomly assign patients to get aprotinin or not. In their analysis, the researchers took into account how sick patients were before surgery, but they acknowledged that some factors they did not account for may have contributed to the extra deaths.

    Aprotinin joins the painkiller Vioxx, drug-coated stents and other drugs and devices where safety concerns arose after the products were on the market.

    (Agencies)