Diabetes medication usage, verified by ARIC investigators,6 was
taken as a valid diagnosis of diabetes. As assessed at exam 1, 2, 3 and
4 and at the 15th annual follow up interview, diagnosis of diabetes
was the outcome event in all survival analyses. Time to event was
the time between exam 1 and the subsequent contact when diabetes
was first reported. Participants lost to follow up were
censored.