Objective   To establish an economical and effective screening method forundiagnosed diabetes in Shanghai.

  Methods   Based on an epidemiological survey of diabetes mellitus in the community populalion in Shanghai, targe-tpopulation was selected to develop risk factors scoring method and logistic regression model.The targe-tpopulation was divided training group and testing group.Logistic regression analysis on the o riginal data of training group was carried out to screen the main risk factors of diabetes mellitus.The score of the variables was determined based on the 10^* Bvalue and the aggregate score and aggregate incldence were used to predict the risk of undiagnosed diabetes.The validity and reliability of two methods were validated and compared based on the original data of testing group.

  Results   When the threshold value was set up at 18, both sensitivity and specificity of risk factors scoring method foridentifying undiagnosed diabetes were 61.33% and 80.40% respectively.When the threshold value was set up at 0.05, both sensitivity and specificity of logistic regression model were foridentifying undiagnosed diabetes were 67.35% and 76.19% respectively.

  Conclusion   The screeningeffect between risk factors scoring method and logistic regression model were no significant.The risk factors scoring method could serve as an effecti and useful tool formass screeningof diabetes mellitus.