Compositional analysis of movement behaviour data

Compositional analysis of movement behaviour data

Since 2017 I’ve been interested in the inter-relationships between different movement behaviours (physical activity, sedentary behaviour, sleep) and how these influence health and development in children and adolescents.

With the collaboration of Dr Dot Dumuid as an expert in compositional data analysis (and others) we have published research which has examined associations between the daily composition of movement behaviours and various outcomes.

Our 2017 paper in IJBNPA showed associations between daily movement compositions, BMI, and cardiorespiratory fitness (CRF) in primary school children. Time spent in MVPA was most strongly associated with BMI and CRF, and time reallocations showed that replacing MVPA with other behaviours had a stronger influence on outcomes than when MVPA replaced other movement behaviours. 

We have also looked at development outcomes such as executive function and motor competence, as well as mental health. Dr Andy Atkin led this collaborative paper using data from the Millennium Cohort Study to examine adolescent time-use compositions and mental health. With a sample of primary and secondary school students we used a similar analysis to investigate the associations between 24-hour movement behaviour compositions and mental health outcomes. In this study we also included executive function outcomes. From these analyses it was clear that the influence of different movement behaviours varies depending on the outcomes of interest. It was also evident that the associations were moderated by age group. Dr Richard Tyler led the motor competence element of this project. In this paper we reported that, relative to other movement behaviours, MVPA had the strongest association overall with motor competence outcomes. Hypothetical reallocations of time from LPA or sleep to MVPA (and vice versa) were associated with the largest positive estimated differences in motor competence outcomes.

We have also started to apply compositional data analysis to movement behaviour data expressed in alternative ways to cutpoint-derived time-use estimates. In this 2022 paper we used 9 acceleration ranges to describe a 24-hour intensity spectrum in a large sample of England youth. We looked at this spectrum compositionally and found that the strongest associations with BMI were from time spent in accelerations in excess of 700 mg, which in intensity terms equates to running type activities. We’re currently looking at characteristics of 24-hour movement behaviours which includes compositional analysis elements. Using the ‘Goldilocks Day’ optimal composition method presented by Dumuid and colleagues our analyses have found that different mental health outcomes have different optimal movement behaviour compositions. We hope to publish this work soon.


Total Page Visits: 511 - Today Page Visits: 1