Nazare et al reported the usefulness of body mass index (BMI) and waist circumference (WC) to know the association with visceral adiposity (VAT) and cardiometabolic risk (CMR) by an international cross-sectional study. The investigators handled 4,109 patients (1,935 women and 2,174 men) from 29 countries, and CMR score was calculated as the sum of the 8 binary subscores. They concluded that WC allowed a further refinement to the single use of BMI for predicting VAT and CMR, respectively. I have some concern on their study outcome.
First, the investigators conducted partial correlation analysis to know the association between BMI (BMI + WC) and VAT or CMR score, stratified by gender in their Table 3. Unfortunately, there is no clear statistical evidence on the advantage of adding WC to BMI for the improvement of partial r 2 with VAT or CMR score. In addition, the explanation rate of the association between BMI (BMI + WC) and CMR score was <20%. Furthermore, gender difference seems small on this association.
Second, the investigators adopted a logistic regression analysis of the combination of BMI and WC for each CMR item by setting BMI and WC as independent variables for each component of CMI and type 2 diabetes mellitus. I understand that there is a discrepancy on the association between each component of CMR and BMI + WC, but CMR scores as the combination of each cardiometabolic component should also be specified by multivariate analysis. In addition, gender difference on the association seems complicated.
Finally, the investigators handled ethnicity and gender with different method. Namely, they used ethnicity as adjusted variable and each gender was stratified for their analysis. I suppose that obesity with different ethnicity should be handled with caution. As the different criteria of central obesity was adopted for the definition of metabolic syndrome, I suppose that there is a limitation of international collaboration study in relation to obesity indexes and CMR only using ethnicity item as an adjusting variable.
Anyway, VAT and CMR are not clinical events and their study design cannot permit causality of the association even for CVD or type 2 diabetes mellitus as the investigators mentioned. Further study is required for confirming their conclusion.