Conclusions
%BF was linearly related to 1/BMI with consistent relationship across Hispanic/Latino backgrounds. BMI cut points consistently underestimated the proportion of Hispanics/Latinos with excess adiposity.
Methods
%BF by bioelectrical impedance analysis was corrected against %BF by (18) O dilution in 434 participants of the ancillary Hispanic Community Health Study/Study of Latinos. Corrected %BF was regressed against 1/BMI in the parent study (n = 15,261), fitting models for each age group, by sex, and Hispanic/Latino background; predicted %BF was then computed for each BMI cut point.
Objective
To evaluate the percentage of body fat (%BF)-BMI relationship, identify %BF levels corresponding to adult BMI cut points, and examine %BF-BMI agreement in a diverse Hispanic/Latino population.
Results
Bioelectrical impedance analysis underestimated %BF by 8.7 ± 0.3% in women and 4.6 ± 0.3% in men (P < 0.0001). The %BF-BMI relationship was nonlinear and linear for 1/BMI. Sex- and age-specific regression parameters between %BF and 1/BMI were consistent across Hispanic/Latino backgrounds (P > 0.05). The precision of the %BF-1/BMI association weakened with increasing age in men but not women. The proportion of participants classified as nonobese by BMI but as having obesity by %BF was generally higher among women and older adults (16.4% in women vs. 12.0% in men aged 50-74 years). Conclusions: %BF was linearly related to 1/BMI with consistent relationship across Hispanic/Latino backgrounds. BMI cut points consistently underestimated the proportion of Hispanics/Latinos with excess adiposity.
Trial registration
ClinicalTrials.gov NCT02060344.