Cohort study to determine the waist circumference cutoffs for predicting type 2 diabetes mellitus in rural China.

Abstract

Limited information is available on the cutoffs of waist circumference (WC) for predicting type 2 diabetes mellitus (T2DM). We aimed to define the optimal WC cutoffs for predicting T2DM among rural Chinese people. A cohort of 11968 participants (732 new-onset T2DM) from a rural area in China with age 18 to 87years was established at baseline during July to August of 2007 and 2008 and followed up during July to August of 2013 and 2014. Scatterplot, X-tile plot, and receiver operating characteristic (ROC) curve analyses were used to determine WC cutoffs for predicting T2DM. The WC cutoffs for males and females were 84 and 86cm (scatterplot), 83 and 88cm (X-tile plot), and 87 and 88cm (ROC curve). According to the highest risk score, the optimal WC cutoffs were 87cm for males and 88cm for females. With the optimal WC cutoffs, the sensitivity, specificity, positive likelihood ratio, area under the ROC curve, and population-attributable risk proportions were 67.9%, 67.0%, 2.06%, 0.70%, and 46%, respectively, for males and 52.5%, 75.0%, 2.10%, 0.69%, and 34%, respectively, for females; the corresponding adjusted hazard ratio for WC predicting T2DM was 3.66 (95% confidence interval 2.80-4.78) for males and 2.55 (2.08-3.12) for females. The optimal WC cutoffs for predicting T2DM were similar between males and females. As well, the criteria of WC for central obesity are no longer practical for predicting T2DM.