Comparative Study Of Predicting The Cerebrovascular Events In Diabetic Patients Between Framingham Risk Score And UKPDS Risk Engine

Objectives: Diabetic cerebrovascular disease is one of the major complicationsof diabetes, and it is an important cause of morbidity and mortality in patients withdiabetes mellitus. Risk prediction models for cerebrovascular events could be greatlyhelpful for the early diagnosis and treatment. Therefore, the aim of our study was tofind the correlation between the different risk factors and cerebral vascular events inpatients with diabetes,and to estimate the predictive performance in predictingcerebrovascular events of Framingham10year Stroke risk score (FSRS) and UKPDSrisk engine.Methods: The1925subjects of our study is from cross sectional analysis ofdiabetes in Meilong community of Minhang District, Shanghai, which started fromMarch to July in2013. Using FRS and UKPDS risk engine to predict the risk ofcerebrovascular events, and analysis the correlation between the different risk factorsand cerebral vascular events by crosstab table, Logistic regression univariate andmultivariate analysis. Comparing similarities and differences of these two models（FSRS and UKPDS risk engine）by crosstab table and variance analysis. Estimatethe predictive performance of FSRS and UKPDS risk score in predicting diabeticcerebral vascular events, by making receiver operating characteristic curve (ROC).Results: To analyze the correlation between the different vascular disease riskfactors and cerebral vascular events, we found that age (or diabetes diagnosis age,duration of diabetes), male, total cholesterol levels, cardiovascular disease history andhistory of hypertension in diabetic patients are the independent risk factors tocerebrovascular events. We set a multiple model including these risk factor which according to the OR value ranging from high to low. In model1, the risk factors arehistory of hypertension (2.202), history of cardiovascular disease (1.537), age (1.062),total cholesterol (0.792), male (0.540); The risk factors of model2are history ofhypertension (2.225), history of cardiovascular disease (1.547), age at DM diagnosis(1.060), diabetes mellitus (1.060), total cholesterol (0.792), male (0.540). Using FSRSand UKPDS risk engine for risk prediction of cerebral vascular events in10years,then making336cases of cerebral vascular events as the gold standard to draw thereceiver operating characteristic curve (ROC curve). To calculate the area under thecurve of the two risk engine, the area under the curve of UKPDS0.582(P<0.001),and area FSRS curve was0.600,(P<0.001).Conclusion: Aging, diabetes, age at diagnosis, gender, duration of diabetes,increased levels of total cholesterol, cardiovascular disease and hypertension riskfactors of cerebral vascular events are independent risk factors. The evaluations ofFramingham10year stroke risk score and UKPDS10year stroke risk engine for10years of cerebrovascular events are meaningful. The predictive performance ofFramingham stroke risk score is slightly better than which of the UKPDS model.