Evaluation of factors determining the precision of body composition measurements by air displacement plethysmography

Objectives: To investigate methodological precision of air displacement plethysmography for assessment of body composition in a heterogenous sample of adults. Design: Accuracy of volume measurements by air displacement plethysmography (ADP) for a range of known volumes was ascertained. Repeated measurements of body volume, lung volume, and derived body composition using the BODPOD measurement system were performed. Influence of surface area estimation on ADP measurement was investigated as a possible source of variation. Setting: Clinical Nutrition Laboratory, School of Health & Sports Science, University of North London, London, UK. Subjects: One hundred and two healthy subjects (57 women, 45 men) who ranged in age between 16 and 55 y and in BMI (kg/m(2)) between 17.8 and 41.9. Study design: Cross-sectional study of healthy adults for comparison with previous studies. Repeat measurements of raw body volume, lung volume and % body fat (BF) by ADP were all performed on the same day. Results: From the range of known volumes a marked increase in the CV and a significantly greater measurement error were found at volumes below 40 l (P = 0.04). Repeat measurements of raw body volume in human subjects resulted in a technical error equivalent to 0.8% BF. There was no significant difference found between measured and predicted lung volume and the 95% confidence interval for difference was only 0.3% BF. Repeat measurements of lung volume in our subset resulted in a technical error equivalent to 0.5% BF. Although body surface area estimation only accounted for variation in % BF of 0.1%, the extent of variation appeared to be governed by leanness (P<0.001). Conclusions: Although ADP retains excellent precision, in practice, repeat measurements of ADP should be performed whenever possible to allow for erroneous volume measurement within one procedure. Protocols for ADP measurement should be created with an awareness of those factors, which may affect measurements. Sponsorships: This study was supported by the University of North London Diversity & Development Fund.