Do people know whether they are overweight? Concordance of self-reported, interviewer-observed, and measured body size

Purpose: To evaluate associations among self-reported, interviewer-observed, and measured body size in a healthcare setting. Methods: A total of 543 adult men and women undergoing colonoscopy were enrolled into a cross-sectional study conducted from 2002 to 2008 at the Johns Hopkins Hospital Outpatient Center. Self-reported and interviewer-observed Stunkard body size figure numbers and measured body mass index (BMI) were collected and evaluated. The body size figures and BMI were categorized as normal weight, overweight, and obese. Results: Correlation between self-reported and interviewer-observed body size figure numbers (r = 0.62) was lower than the correlation between self-reported (r = 0.72) or interviewer-observed (r = 0.84) body size figure number and BMI. Participants underestimated body size by about one figure compared with the interviewers (mean 0.92 ± 1.25). Agreement on normal weight, overweight, and obese between the interviewer-observed body size figures and BMI categories (kappa = 0.40) was higher than for the self-reported body size figures and BMI categories (kappa = 0.23). Among participants who judged themselves in the normal weight category by the figures, 38 and 13 % were overweight and obese, respectively, as measured by BMI. Among participants who judged themselves overweight by the body size figures, 57 % were obese as measured by BMI. Conclusions: Although self-reported body size and measured body size were well correlated, participants underestimated their body size in comparison with interviewers. Many individuals misperceive themselves as normal weight when they are overweight or obese by BMI, which may hinder prevention and control efforts.