A Comparison Of Capability Of Metabolic Syndrome And Prediction Models Or Risk Scores In Prediction Of Type 2 Diabetes

Objective To compare the various prediction models of diabetes and metabolic syndrome on the predictive power of type 2 diabetes.Methods Use the population of multiple metabolic disorders and metabolic syndrome cohort study in Jiangsu Province as the research object, excluding patients with diabetes at baseline, with complete information on the study of a total of 3461 people. Calculated the risk score or the risk probability for each subject using the various prediction models of diabetes,the diagnosed of metabolic syndrome were used National Cholesterol Education Program Adult Treatment Panel Report of the 3rd meeting (NCEP-ATPIII) criteria and the Chinese Diabetes Society on the MS diagnosis of the Chinese proposal (CDS). Calculate and compared the sensitivity, specificity,other diagnostic targets and the area under the ROC curve of each prediction model of diabetes and metabolic syndrome in predicting diabetes.Results (1) Total incidences of type 2 diabetes in follow-up time were 3.7%, 3.3% for men, 3.9% for women. (2) The sensitivity of India score, Thailand score and the metabolic syndrome in prediction of new onset type 2 diabetes were higher than that of the American Diabetes Prediction Model (DPM) and Finland score, but theirs specificity were lower than the latter two. The positive predictive value and positive likelihood ratio of MS and DPM which contain fasting plasma glucose variables were higher than other prediction model which do not contain fasting plasma glucose variables. (3) The area under the ROC curve of MS and DPM were significantly larger than the other prediction model.When fasting plasma glucose was removed, the area under the ROC curve of MS and DPM were significantly decreased and the advantage was no longer obvious.Conclusions The capability of prediction models or risk scores which only include non-biochemical indicators were weaker than metabolic syndrome and the American Diabetes Forecast model in prediction of type 2 diabetes; Second, fasting plasma glucose is the most important part of metabolic syndrome in prediction of diabetes, but it can not completely replace the effect of other components of metabolic syndrome.