No Apparent Progress in Bioelectrical Impedance Accuracy: Validation Against Metabolic Risk and DXA

Bioelectrical impedance (BIA) is quick, easy, and safe when quantifying fat and lean tissue. New BIA models (Tanita BC‐418 MA, abbreviated BIA8) can perform segmental body composition analysis, e.g., estimate %trunkal fatness (%TF). It is not known, however, whether new BIA models can detect metabolic risk factors (MRFs) better than older models (Tanita TBF‐300, abbreviated BIA4). We therefore tested the correlation between MRF and percentage whole‐body fat (%BF) from BIA4 and BIA8 and compared these with the correlation between MRF and dual‐energy X‐ray absorptiometry (DXA, used as gold standard), BMI and waist circumference (WC). The sample consisted of 136 abdominally obese (WC ≥ 88 cm), middle‐aged (30–60 years) women. MRF included fasting blood glucose and insulin; high‐density lipoprotein cholesterol, low‐density lipoprotein cholesterol, and triglycerides; high sensitive C‐reactive protein, plasminogen activator inhibitor‐1 (PAI‐1), and fibrinogen; and alanine transaminase (ALT) liver enzyme. We found that similar to DXA, but in contrast to BMI, neither %BF BIA4 nor %BF BIA8 correlated with blood lipids or ALT. In the segmental analysis of %TF, BIA8 only correlated with inflammatory markers, but not insulin, blood lipids, or ALT liver enzyme (in contrast to WC and %TF DXA). %TF DXA was associated with homeostatic model assessment insulin resistance (HOMA‐IR) independently of WC (P = 0.03), whereas %TF BIA8 was not (P = 0.53). Receiver‐operating characteristic (ROC) curves confirmed that %TF BIA8 did not differ from chance in the detection of insulin resistance (P = 0.26). BIA estimates of fatness were, at best, weakly correlated with obesity‐related risk factors in abdominally obese women, even the new eight‐electrode model. Our data support the continued use of WC and BMI.