ATLANTA (USA TODAY) -- Doctors who used a genetic test to personalize treatment with warfarin, the world's most widely prescribed blood thinner, cut their patients' hospitalization rates by almost a third, researchers said Tuesday.

The study marks the first nationwide attempt to incorporate genetic testing into the routine use of a drug whose effects vary so much from one person to another that it carries a "black box" warning, the government's most urgent safety alert.

Roughly one in five patients are hospitalized for bleeding within six months of starting the drug. Others develop life-threatening blood clots. Doctors must monitor their patients regularly; it may take weeks for them to settle on a safe and effective dose.

"Warfarin is the second-greatest cause of hospitalizations due to drug complications in the country," says Eric Topol, director of the Scripps Translational Science Institute in La Jolla, Calif., who wasn't involved in the study.

Testing was so beneficial that it surprised lead author Robert Epstein, chief medical officer for Medco Health Solutions.

"We were hoping to find a 20% difference," Epstein says. "When we saw it was 30%, we nearly fell off our chairs."

Despite its risks, warfarin, sold as Coumadin and Jantoven, is a mainstay of heart treatment. It's prescribed to 2 million new patients in the USA each year to guard against strokes, second heart attacks and blood clots.

The new findings are the result of a unique partnership between Medco, which administers drug benefits for 65 million beneficiaries, and the Mayo Clinic.

The study involved 896 patients who were tested early for their sensitivity to warfarin, which is governed by two genes. An additional 2,688 patients who weren't tested served as controls. Those who were tested were 31% less likely to land in a hospital bed for any cause and 28% less likely to be hospitalized for bleeding or a clot, Epstein reported at a meeting here of the American College of Cardiology.

Medco and the health plans it serves also will benefit from testing, he says, because safer alternatives to warfarin now being developed are likely to be much more expensive.

Topol and the authors note that the research has drawbacks because of its small size and research design. As a result, the National Heart, Lung, and Blood Institute is sponsoring a much more rigorous trial, led by Stephen Kimmel of the University of Pennsylvania. "We hope to have results within two years," Kimmel says.

Copyright 2010 USA TODAY, a division of Gannett Co. Inc.