Childhood obesity is of major public health concern with significant health, social and economic impacts, having even potential to reverse the increase in longevity that has been observed. We aimed to estimate the instantaneous rate of transition among various states of body mass index (BMI) categories, in children and to evaluate the effect of a vector of explanatory variables–maternal age and education at delivery, type of delivery, sex and birth weight–on the transitions intensities, taking into account that the transitions times are interval censored.

The participants of this study were part of Generation XXI birth cohort, recruited between April 2005 and August 2006, after delivery, during the hospital stay in Porto, Portugal. Follow-up evaluations of the cohort were carried at 4 (between April 2009 and July 2011), 7 (between April 2012 and March 2014), and at 10 years of age (between July 2015 and July 2017) and a total of 4128 participants were included. A multi-state model was implemented describing how a child moved between underweight, normal weight, overweight, and obesity, between the ages 4 and 10 years. Since it was not possible to observe participants continuously throughout the lifespan, as they were assessed only at follow up established visits, the exact times of state-to-state transitions are interval-censored, that is the transition is only known to have occurred within a bounded time interval. Under this constraint, standard multi-state methods cannot be applied and we used a time-homogeneous Markov multi-state model to describe the movement of a child between four states based on BMI.

The maximum likelihood estimate (MLE) and corresponding 95% confidence interval (CI) for each possible instantaneous transitions were calculated. For children in the overweight state, the progression transition to the normal weight state, and for the obese state were similar (0.0099 versus 0.0095). The progression from normal weight state to overweight state was 15 times more likely (0.0061) than the progression from normal weight state to the underweight (0.0004). In underweight children, the progression to normal weight state was the most likely progression rate observed (0.0302). The hazard ratios (HR) from the discrete-time event history models were calculated and statistically significant results were found for sex and maternal education, although only for progressive transitions, i.e. transition to normal weight in those underweight, to overweight in normal weighted and to obesity in children who were overweight. Females were more likely to experience transitions to normal weight among those underweight (HR=1/0.15 [95%CI: 0.02–0.80]=6.67) as in males the most frequent transitions were to overweight and to obesity (HR=1.16 [95% CI: 1.00–1.35]) and HR=1.34 (95% CI: 1.08–1.65), respectively). A similar pattern was observed when analyzing the effect of maternal education as an HR=1.23 (95% CI: 1.01–1.51) for the transition to normal weight status among children who underweight revealed an increase of 23% in the hazard per year of maternal education, opposite effects were observed for the remaining progressive transitions with hazard ratios of 0.98 (95% CI: 0.97–1.01) and 0.96 (95% CI: 0.93–0.98) per year of education, respectively.

We observed that the more likely transition from normal weight status is to overweight state (MLE 0.0061) and once a child was in overweight state, the higher transition probability was to normal weight state (MLE 0.0099). We also find that higher maternal education seems to be protective from unhealthy weight transitions.