Validation of equations using anthropometric and bioelectrical impedance for estimating body composition of the elderly

Resumen

The increase of the elderly population has enhanced the need for studying aging-related issues. In this context, the analysis of morphological alterations occurring with the age has been discussed thoroughly. Evidences point that there are few information on valid methods for estimating body composition of senior citizens in Brazil. Therefore, the objective of this study was to cross-validate equations using either anthropometric or bioelectrical impedance (BIA) data for estimation of body fat (%BF) and of fat-free mass (FFM) in a sample of older individuals from Florianópolis-SC, having the dual energy x-ray absorptiometry (DEXA) as the criterion-measurement. The group was composed by 180 subjects (60 men and 120 women) who participated in four community Groups for the elderly and were systematically randomly selected by a telephone interview, with age ranging from 60 to 81 years. The variables stature, body mass, body circumferences, skinfold thickness, reactance and resistance were measured in the morning at The Sports Center of the Federal University of Santa Catarina. The DEXA evaluation was performed in the afternoon at The Diagnosis Center through Image in Florianópolis-SC. Twenty anthropometric and 8 BIA equations were analyzed for cross-validation. For those equations that estimate body density, the equation of Siri (1961) and the adapted-equation by Deurenberg et al. (1989) were used for conversion into %BF. The analyses were performed with the statistical package SPSS, version 11.5, establishing the level of significance at 5%. The criteria of cross-validation suggested by Lohman (1992) and the graphic dispersion analyses in relation to the mean, as proposed by Bland and Altman (1986) were used. The group presented values for the body mass index (BMI) between 18.4kg.m-2 and 39.3kg.m-2. The mean %BF was of 23.1% (sd=5.8) for men and 37.3% (sd=6.9) in women, varying from 6% to 51.4%. There were no differences among the estimates of the equations converting body density into %BF (p<0.05). The general anthropometric equations developed by Tran and Weltman (1988) and by Deurenberg et al. (1991) showed to be valid for estimating %BF for males, with a standard error of estimate (SEE) ranging from 3.2% and 3.5% and significant correlations with the criterion- measurement (r = 0.78 and 0.74, respectively, p<0.05). For females, the general anthropometric equations of Durnin and Womersley (1974), Tran and Weltman (1988) and Gonçalves specific equation (2004) were valid. These equations presented SEE ranging from 3.3% and 3.5%, with a constant error varying from -1,2% to 1,9%. For the FFM estimation, the equations of Kyle et al. (2001), Dey et al. (2003) and Sun et al. (2003) did not differ from DEXA measures in men (p>0.05), with a constant error varying from -0.7kg to 2.5kg. For women, the equations of Kyle et al. (2001) and Dey et al. (2003) were valid (constant error from 0.3kg to 2.7kg). When using the BMI categories, the equations of Tran and Weltman (1988) and Durnin and Womersley (1974) overestimated the %BF in men with BMI <25kg.m-2. The same pattern was observed with Gonçalves equation (2005) among women, showing that in those specific groups of seniors those equations were not valid for the %BF estimation. The results of the valid BIA equations were not influenced by the BMI categories. Therefore, the valid equations in the present study can be used with the Brazilian elderly population.