Table 3.

Relative Hazard of Cardiovascular Disease After Adjustment for an Unmeasured Confounder in the Prevalent vs Incident Diabetes Cohort^*

Prevalence of unmeasured confounder in prevalent cohort	Prevalence of unmeasured confounder in incident cohort	HR for CVD associated with unmeasured confounder	Adjusted HR† (2.24‡)	95% CI (2.08 to 2.40‡)
0.2	0.1	2.0	1.94	1.80 to 2.08
0.4	0.1	2.0	1.60	1.49 to 1.71
0.6	0.1	2.0	1.40	1.30 to 1.50
0.8	0.1	2.0	1.26	1.17 to 1.35
0.9	0.1	2.0	1.21	1.12 to 1.30
0.2	0.1	3.0	1.80	1.67 to 1.93
0.4	0.1	3.0	1.40	1.30 to 1.50
0.6	0.1	3.0	1.19	1.11 to 1.28
0.8	0.1	3.0	1.06	0.98 to 1.14
0.9	0.1	3.0	1.01	0.94 to 1.08

CVD = cardiovascular disease, HR = hazard ratio, CI = confidence interval

^*A sensitivity analysis was performed to assess the association between diabetes duration and risk of CVD after adjustment for a hypothetical unmeasured confounder by varying the prevalence of the unmeasured confounder (0.2–0.9) in the prevalent diabetes cohort along with the relative hazard for CVD associated with the unmeasured confounder (HR = 2.0–3.0).

^†Adjusted for age, sex, smoking (ever vs never), calendar year, and histories of hyperlipidemia, hypertension, and obesity (BMI ≥ 30).

^‡Observed HR [95% CI] of CVD for the prevalent relative to incident diabetes cohort