Background: To assess the relative metabolic load or risk imposed by fatness, fat mass (FM) is commonly expressed as a proportion of weight (% fat), while central adiposity is assessed using the ratio of triceps (TRI) to subscapular (SUBS) skinfolds. The statistical validity of these indices, defined as independence of the index from its denominator, has received inadequate evaluation.
Objective: To critically examine commonly used obesity indices, and to propose more appropriate approaches.
Design: Cross-sectional studies.
Subjects: In total, 148 infants; 2195 adult men aged 18 y.
Methods: log-log regression analysis was used to explore the relationships between FM and weight (WT) or fat-free mass (FFM) (adults), and between TRI and SUBS (infants and adults).
Results: The simple indices FM/WT, FM/FFM, TRI/SUBS and SUBS/TRI remained related to their denominators, showing their rankings of individuals to be biased by size or fatness. The appropriate power (p) by which to raise the denominator was determined from regression analyses, and differed from unity in all cases. Both skinfold ratios showed a nonlinear relationship with their denominators.
Conclusion: Simple indices for evaluating whole-body or central adiposity are statistically flawed, and remain influenced by size or fatness of the subject. The index FM/WT is conceptually flawed because FM appears in both numerator and denominator, and the index FM/FFM(p) is preferable. Skinfold ratios are also problematic, and to evaluate central adiposity SUBS alone, or SUBS/FFM(p), may represent more appropriate approaches. These issues should be investigated in other studies in order to address possible influences of age, ethnicity, gender and nutritional status.