Cardiovascular disease risk prediction in type 1 diabetes: accounting for the differences

Present analyses used data from the Pittsburgh Epidemiology of Diabetes Complications Study, a prospective study of subjects with childhood type 1 diabetes (T1D), diagnosed between 1950 and 1980. Baseline exams took place 1986-1988 with biennial exams since. The Framingham risk equation was applied to generate the probability of risk for coronary heart disease (CHD) (MI, CHD death, or Q-waves) in 552 CHD free subjects who experienced 42 events over the 10-year follow-up period. Probabilities were split in to deciles. Expected and observed events were compared and demonstrated poor prediction. Risk factors previously found to be associated with CHD in T1D other than those in the Framingham risk function (age, smoking, cholesterol/HDLc, systolic blood pressure) were compared within the highest risk deciles. In men, elevated fibrinogen (p=0.007), white blood cell count (WBC) (p=0.037), albumin excretion rate (AER) (p=0.0001), and lower HDLc (p=0.048) were predictive. In females, higher Beck Depression Inventory (p=0.008), HbA1 (p=0.008), AER (p=0.01), LDLc (p=0.007), fibrinogen (p=0.006), WBC (p=0.005), non-HDLc (p=0.0005), WHR (p=0.003), and estimated glucose disposal rate (p=0.002) were associated. Risk factors not considered by the Framingham risk equation may account for the lack of fit and should be examined further.