Objectives To examine whether anthropometric measures could predict diabetes incidence in a Chinese population during a 15-year follow-up.

Methods The data were collected in 1992 and then again in 2007 from the same group of 687 employees. Waist circumference, body mass index, waist to hip ratio and waist to height ratio were collected based on a standard protocol. To assess the effects of baseline anthropometric measures on the new onset of diabetes, logistic regression models were used to estimate the odds ratios (ORs) of them, and the discriminatory power of anthropometric measures for diabetes was assessed by the area under the receiver operating curve (AROC).

Results Of the 687 individuals without diabetes at baseline, 74 individuals were diagnosed with diabetes during a 15-year follow-up period (incidence 10.8%). These anthropometric measures also predicted future diabetes during a long follow-up (p < 0.01). The areas under the ROC curves were 0.668 (95% CI: 0.601- 0.734) for BMI, 0.701 (95%CI: 0.641-0.760) for WC, 0.691 (95% CI: 0.637-0.748) for WHpR and 0.715 (95% CI: 0.657- 0.774) for WHtR, respectively (all p < 0.001). And differences between AUCs had no significantly differences in four anthropometric measures.

Conclusions These anthropometric measures could predict diabetes with a long time follow up and this might be used in most clinical practices, epidemiological studies and preventive medicine in the future. Future researches may be warranted to assess whether the validity of these predictors are stable in mainland China with different duration of follow-up time in a large population.