RESUMO
OBJECTIVE: Validation of body adiposity index (BAI) in a paediatrics sample; and to develop, if necessary, a valid BAI for paediatrics (i.e. BAIp). METHODS: A total of 1615 children (52% boys) aged 5-12 years underwent anthropometry. Their body composition was assessed using a foot-to-foot bioimpedance. The validity of BAI = (Hip circumference/Height(1.5)) - 18 was tested by combining correlation and agreement statistics. Then, the sample was split into two sub-samples for the construction of BAIp. A regression was used to compute the prediction equation for BAIp-based percentage of body fat (%BF). RESULTS: The initial BAI over-estimated the %BF of children by 49% (29.6 ± 4.2% versus 19.8 ± 6.8%; p < 0.0001). The original methodology led to a BAIp = (Hip circumference/Height(0.8)) - 38 in children. When compared to BAI, BAIp showed both better correlation (r = 0.57; p < 0.01 versus r = 0.74; p < 0.0001) and agreement (ICC = 0.34; [95% CI = -0.19-0.65] versus ICC = 0.83; [95% CI = 0.81-0.84]). However, there were some systematic biases between the two values of %BF as exemplified by the large 95% limit of agreement [-9.1%; 8.8%] obtained. CONCLUSION: BAI over-estimates the %BF in children. In contrast, BAIp appears as a new index for children's body fatness, with acceptable accuracy. In its current form, this index is valid only for large-scale studies.