Prediction of 10-year vascular risk in patients with diabetes: the AD-ON risk score

Aims: To formulate a combined cardiovascular risk score in diabetes that could be useful both to physicians and healthcare funders. Methods: Data were derived from the Action in Diabetes and Vascular Disease: Preterax and Diamicron Modified Release Controlled Evaluation Observational (ADVANCE-ON) study, a randomized controlled trial (mean duration 5 years) with a post-randomization follow-up (mean 4.9 years), that included 11 140 high-risk patients with diabetes. The outcome analysed was the occurrence of either fatal or non-fatal macrovascular or renal disease. A Cox regression model was used to determine weightings in the risk score. The resultant score was recalibrated to each of three major global regions, as covered by the ADVANCE-ON study. Results: Over a median of 9.9 years, 1145 patients experienced at least one component of the combined outcome event. The resultant score, the AD-ON risk score, incorporated 13 demographic or clinical variables. Its discrimination was modest [c-statistic= 0.668 (95% confidence interval 0.651, 0.685)] but its calibration was excellent (predicted and observed risks coincided well, within disparate global regions). In terms of the integrated discrimination improvement index, its performance was marginally superior, over a 10-year risk horizon, to existing risk scores in clinical use, from a restricted version of the same data, for macrovascular and renal disease separately. Conclusions: The AD-ON risk score has advantages over the existing vascular risk scores in diabetes that used data from the original ADVANCE trial, which treat macrovascular and renal diseases separately. These advantages include its simplicity of use and global application.