Body Mass Index, the Most Widely Used but also Widely Criticized Index: Would a Gold-Standard Measure of Total Body Fat be a Better Predictor of Cardiovascular Disease Mortality?

Abstract

Objectives—To examine whether an accurate measure (using a gold-standard method) of total body fat (BF) would be a better predictor of cardiovascular disease (CVD) mortality than body mass index (BMI). Participants and Methods—A total of 60,335 participants were examined between January 1, 1979, and December 31, 2003, and then followed-up for mean of 15.2 years. BMI was estimated by standard procedures. Indices of body composition [i.e. BF%, fat mass index (FMI), fat-free mass (FFM) and FFM index (FFMI)] were derived from either skinfold thicknesses or hydrostatic weighing. For exact comparisons, the indices studied were categorized identically using sexspecific percentiles. Results—Compared with a medium BMI, a very high BMI was associated with hazard ratios (HR) of 2.7 (confidence interval, CI:2.1-3.3) for CVD mortality, a stronger association than for BF % or FMI; i.e. HR=1.6(CI:1.3-1.9) and 2.2(CI:1.8-2.7), respectively. Compared with a medium FFMI, a very high FFMI was associated with a HR of 2.2 (CI:1.7-2.7) for CVD mortality, with these estimates being markedly smaller for FFM, i.e. HR=1.2(CI:0.9-1.6). When the analyses were restricted only to the sample with hydrostatic assessments (N=29,959), the results were nearly identical, with even slightly larger differences in favor of BMI, i.e. HR=3.0 (CI:2.2-4.0) compared with BF% and FMI, i.e. HR=1.5(CI:1.2-1.9) and 2.1(CI:1.6-2.7) respectively. We estimated Harrell c-index as an indicator of discriminant/predictive ability for these models and observed that the c-index in models including BMI was significantly higher than that in models including BF% or FMI (all P values <.005). Conclusions—The simple and inexpensive measure of BMI can be as clinically important or even more than total adiposity measures assessed by accurate and expensive methods. Physiological explanations for these findings are discussed.