Acceptability, Precision and Accuracy of 3D Photonic Scanning for Measurement of Body Shape in a Multi-Ethnic Sample of Children Aged 5-11 Years: The SLIC Study

BACKGROUND: Information on body size and shape is used to interpret many aspects of physiology, including nutritional status, cardio-metabolic risk and lung function. Such data have traditionally been obtained through manual anthropometry, which becomes time-consuming when many measurements are required. 3D photonic scanning (3D-PS) of body surface topography represents an alternative digital technique, previously applied successfully in large studies of adults. The acceptability, precision and accuracy of 3D-PS in young children have not been assessed. METHODS: We attempted to obtain data on girth, width and depth of the chest and waist, and girth of the knee and calf, manually and by 3D-PS in a multi-ethnic sample of 1484 children aged 5–11 years. The rate of 3D-PS success, and reasons for failure, were documented. Precision and accuracy of 3D-PS were assessed relative to manual measurements using the methods of Bland and Altman. RESULTS: Manual measurements were successful in all cases. Although 97.4% of children agreed to undergo 3D-PS, successful scans were only obtained in 70.7% of these. Unsuccessful scans were primarily due to body movement, or inability of the software to extract shape outputs. The odds of scan failure, and the underlying reason, differed by age, size and ethnicity. 3D-PS measurements tended to be greater than those obtained manually (p<0.05), however ranking consistency was high (r(2)>0.90 for most outcomes). CONCLUSIONS: 3D-PS is acceptable in children aged ≥5 years, though with current hardware/software, and body movement artefacts, approximately one third of scans may be unsuccessful. The technique had poorer technical success than manual measurements, and had poorer precision when the measurements were viable. Compared to manual measurements, 3D-PS showed modest average biases but acceptable limits of agreement for large surveys, and little evidence that bias varied substantially with size. Most of the issues we identified could be addressed through further technological development.