Change in clusters of lifestyle behaviours from childhood to adolescence: a longitudinal analysis.

Abstract

This study aimed to identify changes in clusters of lifestyle behaviours (physical activity, screen time and diet) between the ages of 7 and 14years, and to examine socio-demographic determinants of changes. Longitudinal analyses were performed on a sample of 9339 children from the UK Millennium Cohort Study (MCS) who had complete data on behaviours of interest at age 7 (wave 4) and 14years (wave 6). Joint Correspondence Analysis (JCA) and k-means cluster analysis were used to identify clusters of lifestyle behaviours at both time waves. Multinomial logistic regressions were used to examine the associations between socio-economic variables and changes in cluster membership. Analyses were conducted separately for boys and girls. Clusters of behaviours at age 7 and 14years were identified as healthy, mixed or unhealthy respectively. Compared to girls, a higher proportion of boys remained in the healthier cluster over time (19.1% vs. 13.1%) or became healthier (26.4% vs. 9.36%). A higher proportion of girls changed to an unhealthier cluster (57.2% vs. 33.9%). Indicators of lower socio-economic status, such as low family income, low parental education, and not living with both parents at age 7 were associated with unhealthier changes in cluster membership. Conclusion Lifestyle behaviours cluster in children and are susceptible to change over a 7-year period, with a high proportion of boys becoming healthier and a higher proportion of girls became unhealthier. Indicators of socio-economic status appear to be important in determining changes in clusters. What is Known: • Poor lifestyle behaviours (i.e. unhealthy dietary habits, low physical activity, and sedentary behaviours) tend to cluster in children and adolescents. What is New: • Lifestyle behaviours cluster in children and are susceptible to changes between childhood and adolescence. Changes occur differently in boys and girls. Indicators of low socio-economic status are associated with unhealthier changes in behavioural clusters.