RESUMO
This study aims at assessing the accuracy of estimates of body composition provided by bioimpedance (BIA) equations developed for U.S. populations when applied to a sample of Guatemalan farmers. If these equations were shown to have low validity, the second objective was to develop more accurate estimates of fat-free mass (FFM). One hundred males and females 19 to 45 years of age were randomly selected from four rural communities in the Western Highlands of Guatemala. Bioimpedance equations explained 59 and 33% of the variation in FFM, with a RMSE of 2.7 and 2.8 kg in males and females, respectively. Body fat (BF) predictions had a lower R2 . Using the "all possible regressions" procedure, the best subset for prediction of FFM used anthropometric and BIA variables as predictors. The best model for men and women included only anthropometric variables: 75% of the variance in FFM for men and 70% of the variance in women was explained by this model. The RMSE was 2.1 and 1.9 kg for both groups, respectively. It is concluded that FFM can be estimated from anthropometric dimensions with a high degree of accuracy and use of BIA does not provide more valid estimates.