Accuracy of bioelectrical impedance in children and adolescents: systematic review and field study

Bioelectrical impedance analysis (BIA) has been used to investigate body composition in the general population, as well as specific populations such as children and adolescents diagnosed with human immunodeficiency virus (HIV). HIV infection and chronic exposure to antiretroviral therapy may result in reduction of fat-free mass (FFM), low bone mineral content (BMC) and body fat redistribution. The aim of the study was to assess reproducibility, correlation and mean differences in body composition estimated by BIA and reference methods, systematically reviewing the literature in pediatric population, and to determine the validity of body composition analysis by BIA compared to dual-energy X-ray absorptiometry (DXA) and air displacement plethysmography (ADP), in children and adolescents with HIV diagnosis. To identify scientific gaps in the objective of the study, a systematic review was conducted in PubMed, Embase, EBSCO, Web of Science, Scopus and SciELO databases. Subsequently, cross-sectional concurrent validity study was conducted whit sixty-four children and adolescents (8 to 15 years) with HIV diagnosis attended at Hospital-DIA in the Hospital Infantil Joana de Gusmão, Florianópolis, Santa Catarina, Brasil. Fat mass (FM), FFM and percentage fat mass (%FM) were obtained by BIA for comparison with DXA and ADP. In addition, segmented FM (trunk, legs and arms), lean soft tissue mass (LSTM) (total and segmented) and bone mineral content (BMC) were obtained by BIA for comparison with DXA. To determine the validity of body composition analysis by BIA compared to reference methods DXA and ADP the correlation coefficient proposed by Lin, the paired t-test and the Bland-Altman analysis were used. As result of the systematic review, BIA exhibited almost perfect reproducibility for %BF. BIA exhibited almost perfect correlation with the reference methods for FFM and FM, and BIA underestimated FM in children and adolescents. In the field research, BIA presented clinically acceptable correlation with DXA and ADP for FFM, in which values found by ADP were underestimated and values found by DXA were overestimated in children and adolescents with HIV diagnosis. BIA presented clinically acceptable correlation with DXA for LSTM (total and segmented) estimates in both sexes (underestimating FM and overestimating LSTM). BIA presented clinically not acceptable correlation with DXA in %BF, FM and BMC estimation in children and adolescents with HIV diagnosis. It was concluded that %FM estimated by BIA exhibited almost perfect reproducibility in children and adolescents. FM and FFM estimated by BIA correlated almost perfectly with reference methods in both sexes. BIA underestimated FM in both sexes. In addition, BIA was suitable for evaluating FFM and LSTM, and was not suitable for evaluating FM, %BF and BMC in children and adolescents with HIV diagnosis. (author)