Predictive performance of traditional and novel anthropometric indices for diabetes and hypertension

The link between obesity and metabolic dysfunction is well-established. However, the choice of an anthropometric index best reflective of risk remains debatable. This study aimed to evaluate the predictive performance of several indices for diabetes and hypertension in a population at risk for cardiovascular disease.

Data from 1,537 participants was analyzed. The predictive value of 19 indices for diabetes and hypertension was evaluated via area under the receiver operating characteristic curve (AUC) analysis. Analyses were adjusted for major risk factors to evaluate the independent utility of each index. Modified versions of the American Diabetes Association (ADA) diabetes risk assessment tool were examined, where body mass index (BMI) was substituted for indices demonstrating strong or independent predictive values.

The Deurenberg formula was the best predictor of diabetes in both male (AUC = 0.67; 95% CI 0.62-0.73) and female (AUC = 0.77; 95% CI 0.73-0.82) participants, and significantly better than BMI. Body roundness index (BRI; aAUC = 0.63; 95% CI 0.56-0.70), waist-to-height ratio (WHtR; aAUC = 0.63; 95% CI 0.57-0.70), and waist-to-height^1/2 ratio (WHT.5R; aAUC = 0.63; 95% CI 0.57-0.70) showed independent predictive values for diabetes in female participants. The risk assessment tool's performance was improved when BMI was substituted for these indices. BMI (aAUC = 0.66; 95% CI 0.61-0.70), Deurenberg (aAUC = 0.66; 95% CI 0.61-0.70), and Gallagher (aAUC = 0.66; 95% CI 0.62-0.70) formulas were independent predictors of hypertension in male participants.

Several indices showed promising performances for use in diabetes screening. Future research should focus on incorporating these indices in screening tools.