Background: Body mass index (BMI) differs by socio-demographic variables, but the origin of these associations remains relatively unknown.
Objective: To investigate the association between socio-demographic variables and the subsequent change in BMI over six years.
Design: A Dutch prospective cohort study (GLOBE) from which data were used from initially 20-49-year-old subjects (males: n=362; females: n=405). BMI was calculated from self-reported body height and weight data obtained in 1991 and 1997. Socio-demographic variables used were sex, age, educational level and the occupational level of the main breadwinner, family income, marital status, religious affiliation and degree of urbanization and measured in 1991.
Results: Cross-sectionally, BMI was higher in males than in females. BMI was positively associated with age and negatively associated with educational level in both sexes, after adjustment for the other socio-demographic variables. A positive association of BMI with family income was found in males and a negative association with occupational level was found in females. During follow-up, BMI increased significantly more in females (from 23.0 (s.d. 3.3) to 24. 2 (s.d. 3.8)) than in males (from 24.3 (s.d. 2.9) to 25.1 (s.d. 3. 5)). With the exception of a significant lesser increase in BMI in initially 30-39-year-old women compared to initially 40-49-year-old women, no other statistically significant associations were found between socio-demographic variables and the 6-year change in BMI.
Conclusions: Cross-sectional differences in BMI by socio-demographic variables are not due to different 6-year changes in BMI for categories of these variables in adulthood. Cross-sectional differences in BMI by educational level are probably established at the end of adolescence.