Aims/hypothesis: Available multivariable equations for cardiovascular risk assessment in people with diabetes have been derived either from the general population or from populations with diabetes. Their utility and comparative performance in a contemporary group of patients with type 2 diabetes are not well established. The aim of this study was to evaluate the performance of the Framingham and UK Prospective Diabetes Study (UKPDS) risk equations in participants who took part in the Action in Diabetes and Vascular disease: Preterax and Diamicron-MR Controlled Evaluation (ADVANCE) trial.
Methods: The 4-year risks of cardiovascular disease (CVD) and its constituents were estimated using two published Framingham and the UKPDS risk equations in 7,502 individuals with type 2 diabetes without prior known CVD at their enrolment in the trial.
Results: The risk of major CVD was overestimated by 170% (95% CI 146-195%) and 202% (176-231%) using the two Framingham equations. The risk of major coronary heart disease was overestimated by 198% (162-238%) with the UKPDS, and by 146% (117-179%) and 289% (243-341%) with the two different Framingham equations, respectively. The risks of stroke events were also overestimated with the UKPDS and one of the Framingham equations. The ability of these equations to rank risk among ADVANCE participants was modest, with c-statistics ranging from 0.57 to 0.71. Results stratified by sex, treatment allocation and ethnicity were broadly similar.
Conclusions/interpretation: Application of the Framingham and UKPDS risk equations to a contemporary treated group of patients with established type 2 diabetes is likely to substantially overestimate cardiovascular risk.