Using a Recalibrated Cardiovascular Disease Risk Score is Not Adequate to Predict Risk of Recurrence Within 3 Years in Australian Survivors of Stroke

Background: Prognostic performances of models predicting risk of recurrent events of cardiovascular disease (CVD) are not adequate for use in clinical settings. We aimed to determine whether adapting the Framingham Risk Score (FRS) to an Australian population could effectively predict recurrent cardiovascular outcomes. Methods: Patients comprised survivors of stroke/TIA who participated in the Shared Team Approach between Nurses and Doctors For Improved Risk factor Management (STAND FIRM) trial (n = 563). We used standardised anthropometric, biochemical and blood pressure data, collected at baseline, to evaluate risk factors for stroke/TIA. Cox proportional hazards regression models were used to determine the risk of recurrence of CVD-related events and deaths within 3 years after stroke/TIA; adjudicated by two independent stroke specialists. Regression estimates were then used to recalibrate the coefficients used by the FRS, and performance of the model assessed. Results: In women, the recalibrated FRS model had poor discrimination (C-statistic = 0.634) and appeared to better predict CVD recurrence (AUC = 0.664) than the original FRS model (AUC = 0.598). However in men, the recalibrated FRS model had poor discrimination (C-statistic = 0.604) and prediction of CVD recurrence (AUC = 0.632) similar to the original FRS model (AUC = 0.606). Conclusion: The original FRS and recalibrated FRS models appeared to perform poorly in Australian men and women with stroke. The identification of relevant risk factors, easily measured in a clinical setting, may help clinicians better monitor the risks of their patients and enhance secondary prevention strategies.