There has been no analysis to date of the behavioural factors associated with differences in weight gain across socioeconomic strata. Determination of these factors is important for informing public policy that can reduce socioeconomic inequalities in obesity.
We utilised the 21,479 participants from the Melbourne Collaborative Cohort Study who attended the baseline (1990-1994) and follow-up (2003-2007) surveys, were of Australian or Northern European country of birth and were not missing data on education, anthropometric or behavioural factors of interest. We further selected the 5,026 men and 8,671 women who had gained or maintained weight over time (follow up weight ≥ baseline weight). A series of linear regression models were used in accordance with the products of coefficients method to analyse the mediating role of leisure time physical activity, alcohol, soft drink (regular and diet), snacks, savoury items and fruit and vegetable consumption on the relationship between education and body mass index (BMI) at follow-up (adjusted for baseline BMI, representing BMI change). All models were stratified by sex and adjusted for age and smoking status.
Total population mean baseline age was 53, and 43% of these 13,697 persons were of low education.Men showed no significant educational-differences in BMI gain (-0.08 kg/m2 (-0.17, 0.01)). Among women there was significant educational-differences in BMI gain (-0.10 kg/m2 (-0.18, -0.03)) indicating that independent of baseline, women of a high education had a 0.10 kg/m2 lower BMI gain over time. Forty percent of this difference was attributable to two of the mediating factors analysed, alcohol (29%) and diet soft drink (11%).
These results suggest that diet soft drink and alcohol are likely to represent drivers of the relationship between lower education and greater weight gain over time in Australian women. Public policies which target these behaviours may therefore prevent and reduce obesity in an equitable manner.