Giving patients genetic tests to gauge their dose of a widely used blood thinner could reduce the drug’s sometimes deadly side effects, a new study involving University of Wisconsin-Madison researchers suggests.

The drug, warfarin, was discovered at the university in 1948 and is named after the Wisconsin Alumni Research Foundation, or WARF.

About 2 million Americans take warfarin, also known by the brand name Coumadin, and their ideal doses vary greatly. Doctors experiment with different doses and monitor patients with blood tests, but too much of the drug can cause bleeding and too little can fail to prevent blood clots.

Both problems can be fatal. Side effects from warfarin account for more emergency room visits than side effects from any other drug but insulin, according to the U.S. Food and Drug Administration.

Six clinical factors — age, height, weight, race and whether patients are on two other drugs — can help doctors determine the proper dose of warfarin. Variations in two genes, CYP2C9 and VKORC1, also influence a patient’s response to the drug.

In 2007, the FDA announced a new label for warfarin, telling doctors to consider testing of the genes when establishing a dose.

But no large study had shown that genetic testing would indeed lead to better dosing of warfarin — until now.

An international study of about 5,000 patients showed that genetic testing, combined with the six clinical factors, predicted the correct dose of warfarin better than the clinical factors alone. Both techniques were much better than using a standard dose, an approach many doctors follow.

Genetic testing proved most useful in the 46 percent of patients who require especially high or low doses of the drug. Adding the testing to the clinical factors increased those patients’ chances of beginning at or near the right dose by about 50 percent, according to the study, in Thursday’s issue of the New England Journal of Medicine.

"These predictive models give them a much better starting dose," said David Page, a UW-Madison computer scientist who worked on the study. "You avoid problems of overdosing and causing bleeding or under-dosing and allowing clotting to occur."

The study analyzed data from patients already on warfarin.

The National Institutes of Health plans to launch a study next month of patients starting to take the drug, using genetic testing to set the dose for half of the patients and only the clinical factors for the other half. A dozen sites, including the Marshfield Clinic, will participate.

One of the tests for the two genes was developed by Third Wave Technologies of Madison, which was acquired last year by Hologic, of Bedford, Mass.

In 2007, a Third Wave executive said the genetic testing for warfarin would cost health-care payers $100 to $150 but could save more than $1 billion a year by reducing bleeding complications.