Description of study population, N = 420.

Assessment of body composition can be useful in managing many clinical or public health conditions, including HIV. Most people living with HIV infection are in Africa where clinics may lack equipment, utilities or staff time for optimal body composition assessment. Thus, it is important to determine whether less expensive and more available and scalable methods can provide adequate information. We estimated body composition of 420 perinatally HIV-infected Zambian adolescents, aged 11–19 years, using dual-energy X-ray absorptiometry (DXA), bioelectrical impedance (BIA) and anthropometry: body mass index (BMI), waist circumference, and subscapular and suprailiac skinfolds. Data were converted to internal population Z-scores in order to compare measures. BIA and DXA were compared for total fat and fat-free mass, BMI was compared to DXA fat mass, and waist circumference and skinfolds were compared to DXA trunk (central) fat. Neither anthropometry nor BIA adequately reflected total fat or trunk fat as measured by DXA. Although mean bias was generally small, especially for females, the limits of agreement were wide for all comparisons. In addition, for central fat in males, the bias was larger at higher mean Z-score values. BMI had similar agreement with DXA fat mass, as did BIA in females, though not males. We conclude that, because of wide limits of agreement and bias in some measures, none of the simpler methods examined are adequate for assessing longitudinal changes in body composition in order to monitor children’s health. However, where BIA is available and DXA is not, BIA may still be able to describe broad trends in body composition of African adolescents living with HIV.