Benchmarking the Discriminatory Performance of Body Mass Index and Body Fat for Obesity Detection in Patients Treated by Intragastric Balloon.

Abstract

The accuracy of body mass index (BMI) in detecting obesity in patients treated by intragastric balloon (IGB) remains still speculative. We aimed to determine the discriminatory performance of BMI as an estimate of excess body fat (%BF) in an IGB-treated population. Retrospective analysis of prospectively collected data of 476 patients who completed the 6-month IGB treatment period and were subjected to body composition analysis. We evaluated the relationship between BMI and %BF or lean mass and the diagnostic performance of BMI ≥ 30 kg/m2 for adipose tissue detection, stratified by age (< 40 and ≥ 40 years) and gender. Moreover, we identified anthropometric and body composition parameters serving as predictors of obesity according to %BF-based criteria (> 25% in men or > 35% in women). Gender emerged as an effect modifier in the quadratic polynomial relationship between BMI and %BF (R2 = 0.849 for men, R2 = 0.715 for women), while BMI was linearly associated with %BF in both age groups (R2 = 0.435 for men, R2 = 0.474 for women). BMI was strongly correlated with both %BF (r = 0.67) and lean mass (r = - 0.65). The overall area under the ROC curve for BMI ≥ 30 kg/m2 to detect %BF was 0.87 (95%CI 0.85-0.90). A regression model including lean mass, total body water, age, BMI, and female gender explained 0.970 of the variance in %BF. The discriminatory performance of BMI as an estimate of excess body fat is enhanced by the implementation of gender- and age-specific BMI thresholds for defining obesity, in IGB-treated patients.