Long-term trends in BMI: are contemporary childhood BMI growth references appropriate when looking at historical datasets?

Background Body mass index (BMI) is the most widely used surrogate measure of adiposity, and BMI z-scores are often calculated when comparing childhood BMI between populations and population sub-groups. Several growth references are currently used as the basis for calculation of such z-scores, for both contemporary cohorts as well as cohorts born decades ago. Due to the widely acknowledged increases in childhood obesity over recent years it is generally assumed that older birth cohorts would have lower BMIs relative to the current standards. However, this reasonable assumption has not been formally tested. Methods Two growth references (1990 UK and 2000 CDC) are used to calculate BMI z-scores in three historical British national birth cohorts (National Survey of Health and Development (1958), National Child Development Study (1958) and British Cohort Study (1970)). BMI z-scores are obtained for each child at each follow-up age using the lambda-mu-sigma (LMS) method, and their distributions examined. Results Across all three cohorts, median BMI z-score at each follow-up age is observed to be positive in early childhood. This is contrary to what might have been expected given the assumed temporal increase in childhood BMI. However, z-scores then decrease and become negative during adolescence, before increasing once more. Conclusions The differences in BMI distribution between the historical cohorts and the contemporary growth references appear systematic and similar across the cohorts. This might be explained by contemporary reference data describing a faster tempo of weight increase relative to height than observed in older birth cohorts. Comparisons using z-scores over extended periods of time should therefore be interpreted with caution