Introduction: Diabetes is increasing in the US. The USPSTF and ADA recommend screening based on body mass index (BMI). We evaluated if waist circumference was a better predictor of diabetes than BMI.
Methods: We used NHANES (2015-2018). We identified nonpregnant adults ages ≥20 years (N=10,614). A mediation analysis was performed, assessing if the association of BMI (kg/m2) with diabetes prevalence was mediated by waist circumference (cm), using survey weighted logistic regression. Estimates were US representative. Results: We identified 233,874,251 weighted adults. A one-point BMI increase was associated with a 0.95% (95% CI: 0.84%, 1.07%, P<0.001) increase in diabetes prevalence. Similarly, a one-centimeter increase in waist circumference was associated with a 0.51% (95% CI: 0.46%, 0.55%, P<0.001) increase in diabetes prevalence. After adjusting for waist circumference, BMI was not associated with diabetes prevalence (P<0.001) (Figure), while waist circumference continued to predict diabetes prevalence by 0.87% (95% CI: 0.73%, 1.01%, P<0.001) per one-centimeter increase. The association of BMI for diabetes prevalence was fully mediated by waist circumference (indirect effect ≥ total effect).
Conclusion: The predictive ability of BMI for diabetes is fully mediated by waist circumference. Clinical guidelines could consider waist circumference to replace BMI in screening guidelines.
R.Aggarwal: None. N.Chiu: None. D.L.Bhatt: Board Member; Bristol-Myers Squibb Company, Research Support; Lexicon Pharmaceuticals, Inc., Sanofi, Lilly, Boehringer Ingelheim Inc., AstraZeneca, Novo Nordisk.
