Predictive validity of derived clinical adiposity parameters in the assessment of type 2 diabetes mellitus among adults

Abstract

Background: Body Mass Index (BMI) is a standard universal method of assessing clinical adiposity risk. Several studies have shown that obesity as defined by BMI criteria is a risk factor for type 2 diabetes. However, among Asian Indians there is an increasing incidence of thin-fat phenotype where BMI fails as a clinical risk predictor for diabetes. So, there is a need for alternate clinical adiposity parameters. Objective: To assess the predictive validity of derived clinical adiposity parameters in the assessment of Type 2 Diabetes Mellitus among adults attending a tertiary hospital. Methods: A cross-sectional comparative study was performed in which 100 diabetic adult men and women attending the diabetic Out Patient Department were enrolled. As a comparison group, 53 non-diabetic adults attending the OPD and other non-diabetic relatives of patients were enrolled. Anthropometric measurements such as height, weight, waist and hip circumference were measured and derived adiposity parameters such as Body Mass Index, Waist-Hip Ratio, Waist-Height Ratio, Body Adiposity Index and Conicity Index were calculated. Receiver Operating Characteristic curves was plotted for each adiposity parameter with diabetes status as a dependent variable to assess predictive validity of the adiposity parameters in assessing diabetes risk. Age and gender-segregated analysis were also performed. Results: Waist-Hip ratio cut-off=1.009 (AUC=0.956; 95%CI 0.927-9-0.985) sensitivity=90%, specificity=81.1% had the best predictive ability for diabetes. Body adiposity index cut-off=30.5285; (AUC=0.568 95%CI 0.470-0.665) sensitivity=52%, specificity=50.9% was the least predictive. Waist-Height Ratio cut-off=0.5889 (AUC=0.708; 95%CI 0.624-0.792) sensitivity=62%, specificity=62.3% and Conicity Index cut-off=1.5393 (AUC= 0.692; 95% CI 0.603- 0.781) sensitivity=61%, specificity=60.4% had moderate predictive ability and finally Body Mass Index (BMI) cutoff= 26.25 (AUC=0.585; 95% CI 0.491-0.679) sensitivity=56%, specificity=54.7% had a poor predictive ability. Gender segregated analysis showed that BMI and Body Adiposity Index were poor predictors of diabetes in women. Conclusion: Waist-Hip Ratio has the best predictive validity in the assessment of risk of type 2 diabetes. This must be confirmed using longitudinal studies.

Keywords

• obesity
• diabetes
• thin-fat phenotype
• clinical adiposity parameters
• waist-hip ratio