The effect of height on estimates of the change in BMI-based prevalence of childhood obesity

Background/objectives: Body mass index (BMI, body mass/height²) is biased toward height in children. Here we investigate how change in population height affected change in BMI-based estimates of the prevalence of overweight and obesity in Australian children. Subjects/methods: Height, weight, and percent body fat (%BF) were measured at ages 8, 10, and 12 years (1855 sets of measures). Age-specific relationships between BMI and height were derived, adjusting for %BF, to estimate the degree of height bias inherent in BMI. Then, from cross-sectional measurements recorded in 1985 (N = 2388) and 1995 (N = 2148) in 8, 10, and 12 year olds, changes in overweight/obesity prevalences were calculated before and after accounting for the BMI-height bias. Results: Estimates of the effect of height on BMI following adjustment for %BF were similar across age groups and all were significant at p < 0.001. Referring to 12 year olds, at the same %BF for a 1% increase in height there was 0.77% (95% CI 0.55, 0.99) increase in BMI in boys, and 0.74% (0.28, 1.02) increase in girls. Between 1985 and 1995, mean height of 12-year-old boys and girls increased 3.9 and 3.2 cm, respectively. In 1985 unadjusted prevalences of combined overweight/obesity in boys and girls were 13.5% and 13.0%, respectively, and in 1995 were 24% and 24.5%. The latter values were reduced to 21.6% and 22.6% after adjusting for increased height. Conclusions: Previously reported increases in childhood overweight/obesity in Australia between 1985 and 1995 were likely to be moderately overestimated as a result of increased population height; suggesting that population height be taken into account in any pediatric investigation of changes in overweight/obesity prevalence over time.