“The two predictor models were helpful, but discrepancies between them sometimes led to confusion,” says Gage. “We needed a simple, uniform system to help select warfarin for patients at moderate or high risk of stroke, while avoiding this potentially dangerous blood thinner in low-risk patients.”