Continuous-time modeling of the multivariate relationships between physical activity levels and stationary time in preschool-aged children: an investigation of the ActivityStat hypothesis.

BACKGROUND

The ActivityStat hypothesis proposes that an increase or decrease in physical activity (PA) greater than a certain set point activates behavioural and/or energy compensatory responses to maintain a stable level of total energy expenditure. Few studies have tested this hypothesis in children and even fewer have focused on young children. Therefore, the purpose of this study was to investigate the ActivityStat hypothesis by examining the presence and timeframe of the relationships among PA levels and stationary time (ST) in preschool-aged children.

METHODS

A secondary analysis was performed on repeated measurement data (i.e., day-to-day activity) involving 98 preschool-aged children (age: 4.5 ± 0.7 years) in Edmonton, Canada. Participants were asked to wear an ActiGraph wGT3X-BT on the waist for 7 consecutive days to assess PA levels (i.e., light PA [LPA], and moderate-to-vigorous PA[MVPA]) and ST. Bayesian continuous-time structural equation modeling (CT-SEM) was used to examine the relationship between behaviours over time and the timeframe during which these relationships occur.

RESULTS

Each behaviour (i.e., LPA, MVPA, and ST) positively and meaningfully predicted itself at a later time. These relationships persisted up to 0.5 days later, at which point past behaviour no longer meaningfully predicted future behaviour. In contrast, no relationships were observed between the three behaviours.

CONCLUSIONS

This is the first study to investigate the ActivityStat hypothesis using Bayesian CT-SEM in preschool-aged children. When simultaneously taking into account all dynamic relationships suggested by the ActivityStat, the findings fail to support the hypothesis.