To investigate the impact of dynamic change of waist circumference or body mass index (BMI) on type 2 diabetes mellitus (T2DM) populations in a cohort study.

We not only obtained the baseline survey data from program 'Prevention of Multiple Metabolic Disorders and metabolic syndrome (MS) in Jiangsu Province'(PMMJS) which started in 1994, and we conducted twice follow-ups from January 2002 to August 2003, and March 2006 to November 2007. After excluding subjects who were found to have T2DM at baseline, cardiovascular disease(CVD), and BMI<18.5 kg/m² , and loss to follow up because of relocation, death or other reasons, a total of 3 461 subjects were included in this analysis. They received investigation including questionnaires investigation, measurement and laboratory examination. The differences of gender, smoking, alcohol drinking and T2DM family history in different groups were examined using χ²-test, median and inter-quartile range were calculated for TG, and they were examined by rank test. Four equal parts of the differences of waist circumference and BMI were carried out in the COX regression model, to investigate the association between 2 years change of waist circumference or BMI and incidence of T2DM. We also examined the association between BMI and waist circumference modification and incident risk of T2DM in subjects with normal baseline BMI, baseline obese subjects, subjects with normal baseline waist circumference and baseline abdominal obese subjects.

A total of 3 461 participants (1 406 males, 2 055 females) were investigated, including 160 new T2DM cases (60 males, 100 females) who were from between baseline and the second following up. The accumulative incidence was 4.6% (60/3 461). Multivariate COX regression model analysis results showed that the T2DM risk was relatively high in the highest quartile of waist circumference D-value group(HR=2.06, 95% CI: 1.27-3.16), the T2DM risk was also high in the highest quartile of BMI D-value group (HR=1.30, 95% CI: 0.86-1.95). In subjects with abdominal obesity and normal waist circumference at baseline, the incidence rate of T2DM in non-control group was 7.1% (40/565) , 6.3% (45/645), higher than that in control group (3.4%(71/2 096), 4.5%(4/155)) (χ² values were 3.98 and 15.18, P values were 0.043 and <0.001). In subjects with normal waist circumference, T2DM risk was higher in non-control group than that in control group (HR=2.12, 95% CI: 1.40-3.22). In abdominal obese subjects, T2DM risk was also higher in non-control group than that in control group (HR=1.14, 95% CI: 1.04-1.92). If waist circumference was not controlled, T2DM risk was high, no matter BMI controlled or not (HR(95% CI) were 1.73(1.17-2.54),2.45(1.63-3.69) respectively).

Controlling the waistline could reduce the risk of diabetes, and once waist circumference was not controlled, T2DM risk would be increased no matter BMI was controlled or not.

Waist circumference; Diabetes mellitus, type 2; Prospective study; Body mass index