To assess for potential sample-selection bias (i.e., the possibility that the survey cohort was not representative of the original study population), we sought and were granted a Health Insurance Portability and Accountability Act waiver to ex- amine our primary and secondary outcomes in the complete cohort (see the Supplementary Ap- pendix). In the case of patients who were not in the survey cohort and who had not had an event during the interventional phase of the study, we estimate that only approximately 6% of cardio- vascular outcomes were missing from registry data (specifically, amputations for ischemic gan- grene [events that require adjudication] and strokes and heart attacks in participants younger than 65 years of age at non-VA facilities [since persons younger than 65 years of age are not included in CMS files]).