Evaluating the association of anthropometric indices and their principal components with type 2 diabetes in a...

Type 2 diabetes mellitus is a growing global health concern, particularly in Iran. Body fat distribution is a more balanced indicator of type 2 diabetes mellitus (T2DM) risk than overall weight.

This cross-sectional study investigated the association between anthropometric indices and T2DM risk in 3,916 Iranian adults (35-70 years) from the Bandare-Kong Non-Communicable Disease (BKNCD) cohort. A principal component analysis (PCA) on nine anthropometric measurements: Body shape index (ABSI), body roundness index (BRI), body adiposity index (BAI), waist-to-height ratio (WHTR), waist-to-hip ratio (WHR), Waist Circumference (Waist), body mass index (BMI), hip circumference (Hip), wrist ratio (WR) was conducted. Principal components (PCs) were derived from this analysis. The association between anthropometric indices (and their principal components) and T2DM was assessed using logistic regression. Additionally, we calculated the area under the receiver operating characteristic curve (AUC) and used Youden's J index to identify optimal sex-specific cut-off points for the indices.

In this study involving 3,916 participants, we found that 732 (18.7%) were diabetic, with notable gender differences in anthropometric indices related to diabetes risk. Principal component analysis revealed that for males, the first principal component (PC1) was significantly associated with WHTR, BRI and waist circumference while for females, WHR was particularly important in PC2. Logistic regression analysis indicated positive associations between ABSI, BRI, WHTR, WHR, and Waist with diabetes risk, with odds ratios exceeding 1.0 and p-values less than 0.001. WHR showed the most notable association in females (OR = 1.94) and males (OR = 1.61).WHR demonstrated the best overall discriminative performance for both sexes (AUC: 0.66 in males, 0.72 in females) and was the strongest independent risk factor in females (OR = 1.94). For males, waist circumference showed the strongest association in regression models (OR = 1.50) and high sensitivity (0.84), but its low specificity (0.31) limits its use as a standalone screening tool. Other indices like BRI and WHTR showed significant but more modest associations.

Our results showed that central adiposity indices, especially WHR, are more useful than BMI in identifying T2DM risk in this population,but their diagnostic accuracy is only moderate. The optimal index and its interpretation differ by sex, highlighting the need for sex-specific clinical assessment.