The links between behavioural characteristics and growth trajectories in childhood are unclear. We investigated whether screen time, physical activity, sleep duration, appetitive traits and eating behaviours were associated with BMI trajectory in the GUSTO (Growing Up in Singapore Towards healthy Outcomes) cohort.


BMI trajectories were constructed by latent class growth mixture modelling (N=1083) from quarterly (birth-24 mo.) or semiannual (36-72 mo.) anthropometry. Child appetitive traits (3, 12, 36, 60, 72 mo.), sleep patterns (6, 9, 12, 18 mo.), screen viewing and outdoor activities (24, 36 mo.) were measured by parental questionnaire (N=414-932). Eating behaviours (e.g. eating rates and energy intake) were measured in the laboratory (54, 72 mo.; N=382-509). Accelerometry was taken at 66 mo. (N=587). BMI trajectory-child behaviour associations were assessed cross-sectionally, and longitudinally in linear mixed models, with trajectory*time interactions, within-subject random slopes and intercepts.


Two accelerating and 3 stable trajectories were identified. Of assessed behaviours, only appetitive traits and eating behaviours differed. From 12 mo., appetitive traits mirrored BMI trajectories, e.g. children with accelerating BMI had higher slowness in eating at 12 mo. (mean± SE= 2.59± 0.14 vs. 2.47± 0.06 in Stable Normal (SN) trajectory), which reversed by 36 mo. (2.75± 0.10 vs. 3.03± 0.04). By 72 mo. accelerating trajectories had the highest food responsiveness and enjoyment, and the lowest satiety responsiveness and slowness in eating scores (all p<0.001). Eating behaviours reflected appetite differences, e.g. higher eating rate (15.95g/min± 0.96) and energy intake (251.5 kcal± 20.2) in accelerating vs. SN trajectory (12.33g/min± 0.43; 186.6kcal± 9.03; p<0.005).


Appetitive traits over time recapitulated differences in BMI trajectories. Further studies on individual level BMI accounting for temporal sequence will investigate causality and targets for intervention.