Epidemiology characteristic and verification of 14 non-invasive risk score models for screening patients with diabetes
PAN Bing-ying;CHEN Xiong-fei;WU Xue-qi;DONG Xiao-mei;Guangzhou Center for Disease Control and Prevention;
Objective To summarize the characteristics of the published non-invasive risk score models for screening the undiagnosed patients with diabetes to verify in residents of Guangzhou and provide the base for establishing the risk score models in residents of Guangzhou. Methods The literatures were searched for selecting the non-invasive risk score models and the epidemiological characteristics and the results of diagnostic tests of those models were summarized. The data of resident in Guangzhou were used to verify the repeatability of models in resident in Guangzhou. Results Thirteen articles and 14 models were included in the study. Eight predictive indexes in those models were age, obesity-related indicators, gender, family history of diabetes, history of hypertension, sports-related indicators, smoking and the history of using steroids. The minimum, maximum and median values of Youden index were 1.208, 1.521 and 1.400, respectively. The minimum, maximum and median values of area under the receiver operating characteristic curve( AUC) were 0. 673, 0. 884 and 0. 756, respectively. As compared with the literaturess, 2 AUCs increased and 12 AUCs decreased in data of resident in Guangzhou. As compare with literatures, 12 Youden indexes reduced and one Youden index enhanced when using the original cut-off; 11 Youden indexes reduced and 3 Youden indexes enhanced when using the optimal cut-off. Conclusion The most designs of the 14 non-invasive risk score models in screening for undiagnosed diabetes from 13 articles are reasonable and whose results are reliable. Performance in resident in Guangzhou is acceptable. Those models are useful for establishing the risk score models in residents of Guangzhou.