The clustering of health-related behaviours in a British population sample: Testing for cohort differences.

Research findings indicate that health-related behaviours (HRBs) do not co-occur within individuals by chance and therefore cluster. This study uses Latent Profile Analysis (LPA), to identify the clustered patterns and prevalence of four HRBs: smoking, alcohol, diet, physical activity. We used data, collected from participants in their early 30s, from two British cohorts born in 1958 and 1970 (N=21,019). Multi-group LPA models were run separately by gender testing for cohort differences in HRB cluster patterns. For both genders three clusters emerged: 'Risky' (1-9%), 'Moderate Smokers' (20-30%) and 'Mainstream' (68-77%). HRBs amongst members of the 'Mainstream' cluster were more beneficial than HRBs amongst members of the other two clusters, characterised as not smoking, frequent fruit and vegetable consumption, less frequent consumption of chips and fried food and being more physically active. Nevertheless, frequent consumption of sweet foods was common in the 'Mainstream' cluster. There was a large shift in membership to the 'Mainstream' cluster for men and women born in 1970. Amongst women members of the 'Mainstream' cluster, a higher proportion of those born in 1970 appeared to have drunk alcohol above the contemporaneous UK recommended limits but consumed sweet foods less frequently, than those born in 1958. In summary our findings provide additional evidence of HRB clustering, identifying largely consistent HRBs cluster patterns across cohort and gender groups, with some differences in prevalence. This evidence of HRB clustering across time and by gender provides a person-centred understanding that can inform interventions to improve HRBs.