How does one measure "double the risk"? Does that mean if one person in a control group happened to have a stroke in that time period, and now it's two people, that counts as double? Out of 4000? Surely that can't be right.