The objective of the present study was to explore the consistency of dietary choices made by children as they grow up. The dietary habits of 342 healthy children were reported annually from 5 to 13 years on a forty-five-item FFQ and analysed by factor analysis. The same two principal dietary patterns – ‘Healthy’ and ‘Unhealthy’ – emerged each year, and their consistency was assessed using Tucker's congruence coefficient (φ). Individual dietary z-scores for both of these patterns were then calculated every year for each child, and their consistency was measured by Pearson's correlation coefficient (r). Linear mixed-effects modelling was used to investigate individual trends and to quantify reliability of the individual dietary z-scores. Dietary patterns were moderately consistent and systematic over time (0·65 ≤ φHealthy ≤ 0·76; 0·62 ≤ φUnhealthy ≤ 0·78). Individual choices were also consistent year-on-year (0·64 ≤ rHealthy ≤ 0·71; 0·57 ≤ rUnhealthy ≤ 0·68). Reliability rose from 70 % with a single measure to over 90 % with four consecutive measures. The quality of diet diminished over time in 29 % of the children and improved in only 14 %. Dietary habits appear to be set early and seldom improve spontaneously.

Foot-to-foot bioelectrical impedance analysis (BIA) is simple and non-invasive, making it particularly suitable for use in children. There is insufficient evidence of the validity of foot-to-foot BIA compared with dual-energy X-ray absorptiometry (DEXA) as the criterion method in healthy young children. Our objective was to assess the validity of foot-to-foot BIA against DEXA in a large cohort of healthy young children. Body composition was measured by foot-to-foot BIA and DEXA in 203 children (mean age 8·9 (sd 0·3) years). Bland–Altman and simple linear regression analyses were used to determine agreement between methods. BIA overestimated fat-free mass by a mean of 2·4 % in boys and 5·7 % in girls, while fat mass was underestimated by 6·5 % in boys and 10·3 % in girls. The percentage fat recorded by BIA was, accordingly, also lower than by DEXA (boys 4·8 %; girls 12·8 %). In boys, however, there were correlations between the size of the difference between methods and the size of the measure under consideration such that in smaller boys fat-free mass was underestimated (r − 0·57; P < 0·001) while fat mass and percentage fat were overestimated (r 0·74 for fat mass; r 0·69 for percentage fat; both P < 0·001) with the reverse in bigger boys. Mean differences between techniques were greater in the girls than in the boys but in boys only, the direction of the differences was dependent upon the size of the child. Therefore, BIA may be useful for large-scale studies but is not interchangeable with DEXA and should be interpreted with caution in individuals.