Table 2

Proportional hazards models of the effect of time-dependent covariates on the risk of CVD and of the effect of the treatment group after adjustment for the time-dependent covariate

Effect of time-dependent covariate*Treatment group adjusted for specific time-dependent covariate*
Time-dependent covariateHR (95% CI)P valueHR (95% CI)P value
None0.70 (0.52, 0.93)0.0151
Sustained microalbuminuria (yes vs. no)2.28 (1.67, 3.11)<0.00010.78 (0.58, 1.05)0.1038
Macroalbuminuria (yes vs. no)§2.22 (1.46, 3.38)0.00020.75 (0.56, 1.01)0.0594
Kidney failure (yes vs. no)4.84 (1.93, 12.11)0.00080.71 (0.53, 0.95)0.0199
Mean HbA1c during DCCT/EDIC#
 Per 10% increase1.35 (1.21, 1.51)<0.00011.00 (0.72, 1.39)0.9843
 Per 10% decrease0.72 (0.63, 0.81)
  • *Each of the above five models was also adjusted for the following DCCT baseline characteristics: HbA1c value, age, cholesterol level, and smoking status.

  • ‡Sustained microalbuminuria was defined by a history of microalbuminuria (AER ≥30 mg/24 h) on at least two consecutive visits, or ESRD (dialysis or transplant).

  • §Macroalbuminuria was defined by a history of albuminuria (AER ≥300 mg/24 h) or ESRD.

  • ¶Kidney failure was defined by a history of eGFR <15 mL/min/1.73 m2 or ESRD.

  • #The weighted mean HbA1c up to the time of each CVD event was calculated as the weighted mean of the quarterly values during DCCT plus the annual values during EDIC, weighted respectively by ¼ and 1 to reflect the interval between values during the DCCT and EDIC. The log mean HbA1c value was used so that the HR per c-fold change in risk is c3.17155, where 3.17155 is the estimated regression coefficient; a c of 1.1 corresponds to a 10% increase in the mean glycosylated hemoglobin value, and a c of 0.9 to a 10% decrease.