Childhood intelligence, educational attainment and adult body mass index: findings from a prospective cohort and within sibling-pairs analysis.

BACKGROUND: The mechanisms underlying the observed association of childhood intelligence with body mass index (BMI) are unclear and few studies of this association have been prospective in design. METHODS: Prospective study in a birth cohort of 5467 individuals who were born in Aberdeen, Scotland between 1950 and 1956 and who responded to a follow-up survey in 2001. Comparison of associations within sibling pairs of the same family to associations between different families in 643 sibling pairs (1286 individuals) who are participants in the main cohort. RESULTS: Childhood intelligence (age 7 years) and educational attainment were both inversely associated with adult BMI (mean age 48 years): the sex- and age-adjusted mean change in adult BMI per s.d. of intelligence was -0.35 kg/m(2) (95% CI: -0.49, -0.21 kg/m(2)) and per unit increase in educational category (seven categories) was -0.28 kg/m(2) (95% CI: -0.34, -0.22). On adjustment for education the association between childhood intelligence and adult BMI attenuated to the null (-0.03 kg/m(2) (-0.19, 0.13 kg/m(2))); other potential confounding or mediating factors had little or only modest effects on this association. The association between education and adult BMI was not affected by adjustment for childhood intelligence or other potential covariates. The within sibling-pair effect of education on adult BMI (-0.06 kg/m(2) (95% CI: -0.26, 0.14)) was weaker than the effect between different families (-0.37 kg/m(2) (95%CI: -0.58, -0.17)), P-value for difference of within sibling and between family effect=0.03. CONCLUSIONS: The association of childhood intelligence with adult BMI is attenuated to the null on adjustment for educational attainment, whereas the association of educational attainment with adult BMI appears to be independent of childhood intelligence and other measured covariates. However, our family analyses suggest that fixed family and neighbourhood factors, which are closely matched in siblings of a similar age, explain much of the association between greater educational attainment and lower adult BMI.