Abstract P139: Association Between Anthropometric Measurements and Diabetes and Their Abilities to Predict Diabetes: Based on the UK Whitehall II Study

Introduction: It is recognized that people with a high body mass index (BMI) will have a high risk of developing diabetes. However, some studies showed that other indices may provide a more accurate insight into the risk of developing diabetes. Our study aimed to compare the predictive abilities of 5 anthropometric measurements BMI, waist circumference (WC), waist-to-height ratio (WHtR), waist-by-height0.5 (WHT.5R), and a body shape index (ABSI) for diabetes. Hypothesis: WHtR may be most strongly associated with diabetes. High WHtR but normal BMI persons may have higher risk for developing diabetes than persons with high BMI but normal WHtR. Methods: From the UK Whitehall II cohort, we included participants with full records of weight, height, and WC in Phase 3 (1991-1994) for baseline survey, and carried out follow-up survey in Phase 9 (2007-2009) for the non-diabetics of the baseline survey. Our study indices, BMI, WC, WHtR, WHT.5R, and ABSI, were calculated from measured weight, height, or WC. Based on the recognized cut-off value of BMI, WC, and WHtR, we classified the persons into normal group (BMI: <25 kg/m 2 ; WC: <90 cm for men and <80 cm for women; WHtR: <0.5) and abnormal group (≥ corresponding cut-off value). For the indices (WHT.5R and ABSI) without an accepted cut-off value, we selected the 75th cohort-wide centile as the cut point to divide the participants into normal group (WHT.5R <75th cohort-wide centile; ABSI <75th cohort-wide centile) and abnormal group (≥ corresponding cut-off value). Subgroup analyses were based on combination of BMI and WHtR (normal BMI + normal WHtR; high BMI + normal WHtR; normal BMI + high WHtR; high BMI + high WHtR). The outcome is diabetes. We studied predictive value of the indices to diabetes by receiver operator characteristic (ROC) curve analysis at baseline survey, and evaluated the association between the indices and diabetes by Cox regression analysis at follow-up survey. Results: A total of 7979 participants were included at baseline survey mean age: 50.1 ± 6.0 years; 2468 (30.9%) females, and there were 7488 non-diabetics available for the follow-up survey. At baseline survey, ABSI areas under the curve (AUC): 0.711, sensitivity: 77.0%, specificity: 55.5% and WHtR (AUC: 0.709, sensitivity: 67.2%, specificity: 65.6%) had top 2 predictive value of diabetes. At follow-up survey, the abnormal group divided by cut-off value of WHtR had the highest hazard ratio (HR) (2.46; 95% CI: 2.06-2.93), while the abnormal group classified by ABSI cut-off value had the lowest HR (1.77; 95% CI: 1.45-2.15). In subgroup analyses, “normal BMI + high WHtR” group (HR: 2.28; 95% CI: 1.59-3.27) appears to have higher risk for developing diabetes than “high BMI + normal WHtR” group (HR: 1.63; 95% CI: 1.22-2.17). Conclusion: WHtR seems to be an effective indicator for predicting diabetes. Besides BMI, controlling WHtR to the normal range is important for the prevention of diabetes.