. "By comparing variables, including weather, between the day of the stroke and a day one week before, each patient served as his or her own control." . . .