Fatness and body mass index from birth to young adulthood in a rural Guatemalan population.

Body mass index (BMI; wt in kg/ht2 in m) has been proposed as a simple and valid measure for monitoring fatness. Using data from a 25-y longitudinal study of rural Guatemalans, we found that, as children, this population was stunted (mean height-for-age z = -2.6) and had low triceps skinfold thicknesses ( approximately 10% of reference medians), yet had mean BMIs above US reference medians. As young adults, mean BMIs were at the 50th and 20th percentiles for women and men, respectively. BMIs between ages 1 and 5 y were moderately correlated (r = 0.2-0.3) with those in young adulthood. BMI was correlated with subscapular (r = 0.5-0.8) and triceps (r = 0.2-0.7) skinfold thicknesses at all ages and with predicted percentage body fat in adolescence (r = 0.65) and adulthood (r = 0.8). Fatness was highly centralized, with ratios of subscapular to triceps skinfold thicknesses at the 50th-90th percentiles of reference medians at all ages. BMI was a poor indicator of central fat; the correlation between BMI and waist-to-hip ratio in 14-17-y-old males was -0.21). In stunted populations in developing countries, BMI alone should be interpreted with caution. In stunted children, BMIs may be high despite small extremity skinfold thicknesses; BMI alone may overestimate the prevalence of fatness in these children. In adults, measures in addition to BMI may be required to identify centralized adiposity in these populations.