There may be two distinct child obesity epidemics - one among infants and one among adolescents - research suggests.

Researchers came to this conclusion after comparing obesity data in the 1980s with that of more than 300 present-day children.

They found signs that obesity was linked to the influence of parents for young children, the International Journal of Obesity reported.

Meanwhile, for teenagers more general peer-group factors seemed to be key.

The researchers said this suggested different approaches may be needed to tackle the problem.

They came to this conclusion after their data suggested the rise in obesity among the very young had largely been restricted to the minority with obese parents.

Rates among toddlers with non-obese parents had hardly changed.

By contrast, obesity among 16-year-olds had risen across the entire age group.

Background/objective: The prevalence of childhood obesity continues to rise in most countries, but the exposures responsible remain unclear. The shape of the BMI distribution curve defines how a population responds, and can be described by its three parameters - skew (L), median (M) and variance (S). We used LMS analysis to explore differences in the BMI trajectories of contemporary UK children with those of 25 years ago, and to draw inferences on the exposures responsible.

Subjects/methods: We applied Cole’s LMS method to compare the BMI trajectories of 307 UK children (EarlyBird cohort) measured annually from 5-16y (2000-2012) with those of the BMI data set used to construct the UK 1990 growth centiles, and used group-based trajectory modelling (GBTM) to establish whether categorical trajectories emerged.

Results: Gender-specific birth weights were normally distributed and similar between both data sets. The skew and variance established by 5y in the 1990 children remained stable during the remainder of their childhood, but the pattern was different for children 25 years on. The skew at 5y among the EarlyBird children was greatly exaggerated, and involved selectively the offspring of obese parents, but returned to 1990 levels by puberty. As the skew diminished, so the variance in BMI rose sharply. The median BMI of the EarlyBird children differed little from that of 1990 before puberty, but diverged from it as the variance rose. GBTM uncovered four groups with distinct trajectories, which were related to parental obesity.

Conclusions: There appear to be two distinct environmental interactions with body mass among contemporary children, the one operating selectively according to parental BMI during early childhood, the second more generally in puberty