After examining the frequency distribution of all variables, the associations between the proteinuria groups and possible confounding variables were assessed by χ2 analysis and analysis of variance. Mortality rates were expressed as the number of deaths from cardiovascular disease per 1000 person-years accrued for each cohort member. This was based on the length of follow-up, calculated as the number of days from the date of the 1984-1986 examination to the date of death, date of last contact, or December 31, 1996, whichever was earliest. The relation of proteinuria and subsequent mortality from cardiovascular disease was examined with Kaplan-Meier analysis.50 To evaluate whether mortality differed by proteinuria groups, the log-rank test51 was used. Multivariate analysis was performed using Cox proportional hazards regression,52 with 2 indicator variables created for the groups with microalbuminuria and gross proteinuria, which compared them with the group with normoalbuminuria. Initial analyses involved adjustments for age and sex, and subsequent procedures further controlled for diabetes-related variables, other cardiovascular risk factors, and presence of comorbid conditions or their markers. The diabetes-related variables we examined included diabetes duration, insulin intake, oral glucose-lowering agent use, plasma C peptide levels (categorized as <0.03, 0.03-0.29, and ≥0.30 nmol/L), and glycemic control (grouped as "excellent," "good," and "take action" based on glycosylated hemoglobin values of <7.5, 7.5-8.6, and ≥8.7%, respectively, which corresponded to the current recommendations for glycemic control using hemoglobin A1c).45,53 Cardiovascular risk factors included cigarette smoking (classified as never, former, current), physical activity (categorized as engaging in regular physical activity for >3, 1-3, or 0 times per week), education (<12, 12, >12 years of school completed), alcohol intake (classified as never drinkers, past-year nondrinkers, and drinkers with alcohol intake of <2 g/d, 2-13 g/d, and ≥14 g/d), body mass index (calculated as weight in kilograms divided by the square of the height in meters), systolic and diastolic blood pressures, history of hypertension (defined as systolic blood pressure of ≥160 mm Hg or a diastolic of ≥95 mm Hg or taking antihypertensive medications), and intake of medications such as aspirin and antihypertensive agents. The comorbid conditions (or their markers) we examined included the presence of peripheral neuropathy symptoms (defined as self-reported loss of tactile sensation in hands or feet or decreased ability to feel the hotness or coldness of things), the presence and severity of diabetic retinopathy (grouped into none, mild to early nonproliferative, moderate to severe nonproliferative, and proliferative retinopathy, based on fundus photographs graded in masked fashion using a modified Airlie House classification system),36,54,55 and history of cardiovascular disease (defined as having prior angina, myocardial infarction or stroke, or intake of nitroglycerin or digoxin). Variables were successively entered in Cox regression models, which adjusted for age, sex, and glycemic control, starting from diabetes-related variables, to other cardiovascular risk factors, and finally, to comorbid conditions or complications, while retaining those that remained independently related to cardiovascular mortality. For a subset (n=400) of the total study sample for whom we had measurements on HDL-C, total cholesterol, and the ratio of total cholesterol to HDL-C, we repeated our multivariate analyses including each of these variables.