Clinically-assessed vs. self-reported measures of BMI: Implications for type 2 diabetes prevalence in U.S. middle-aged and older adults

Objectives: In response to concerns about the accuracy of self-reported (SR) body mass index (BMI; kg/m2) and type 2 diabetes in middle-aged and older adults, particularly older adults who experience more physical and cognitive changes that may affect their self-reports, we examined the accuracy of these measures in adults aged 50-74 and over age 74. Methods: We used contemporary, nationally representative National Health and Nutrition Examination Survey (NHANES) 2005-2006 data (n=713) to study the accuracy of SR BMI (based on SR weight and height) and SR diabetes (SR of doctor diagnosis) compared to CA BMI (based on NHANES clinical height and weight) and CA diabetes (≥126 mg/dL) in two adult age groups, controlling for theoretically and empirically important sociodemographic variables. Results: We found strong positive correlations between SR BMI and CA BMI in both younger and older age groups (r=0.9471, p<0.0001; r=0.9391, p<.0001, respectively) and between SR diabetes and CA diabetes in both groups (r=0.6222, p<0.0001; r=0.6591, p<0.0001, respectively). Logistic regression analyses indicated that the variance in diabetes explained SR BMI in both age groups (R2=0.0846, p<0.0001; R2=0.1466, p<0.0001) were similar to that explained by CA BMI (R2=0.0805, p<0.0001; R2=0.1628, p<0.0001). The same was true when SR and CA BMI predicted CA diabetes. Conclusions: Overweight, underweight, and diabetes are troublesome conditions in aging Americans. Our results indicate that SR measures of these conditions are valid, even in older adults, ensuring accurate and equitable screenings for and research on these chronic conditions in older adults.