Background Body mass index (BMI) and obesity rates have increased sharply in recent decades. Prior studies have highlighted the potential contributing role of cognitive ability on BMI. However, few studies have investigated the association between early life cognition and BMI across the entirety of adulthood and also the changes in these associations over calendar time (i.e. cohort effects). Investigating changes over time is important—an increasing strength of association for example could suggest a need to simplify information provided to maintain or lose body weight. This study investigates the associations between childhood and adolescent cognitive ability and BMI at multiple ages across adulthood, using three British birth cohorts to make cross-cohort comparisons.

Methods Three nationally representative British birth cohorts born in 1946, 1958 and 1970 were used (N=40,058). Cognitive ability was assessed at ages 10/11 and 15/16 using verbal reasoning/vocabulary and mathematical tests. BMI at multiple ages across adulthood were investigated (1946 cohort: ages 20/26/36/43/53/63/69; 1958 cohort: 23/33/42/45/50/55; 1970 cohort: 26/29/34/42/46). Multivariable regression models were used, regressing BMI at each age on cognitive ability, adjusting for childhood socioeconomic factors, parental BMI, and birth weight. Future iterations will also use polygenic scores as an additional measure for cognitive ability and analyses of the 1946 cohort are being conducted.

Results In the 1958 and 1970 cohorts, higher childhood/adolescent cognitive ability was associated with lower adult BMI, with the associations strengthening as age increased. For example, in the 1958 cohort, one standard deviation (SD) increase in age 11 verbal reasoning score was associated with -0.30 (95% confidence interval (CI): -0.36, -0.25) unit lower BMI at age 23 and -0.51 (-0.62, -0.40) lower at age 55. After adjusting for potential confounders, associations persisted across all ages in the 1958 cohort while for the 1970 cohort only the associations after mid-adulthood (at age 42 and 46) remained. However, the associations for the 1970 cohort at age 42/46 were greater compared with the 1950 cohort at similar ages, suggesting that the pattern of the associations across ages differs between cohorts.

Conclusion Higher childhood/adolescent cognitive ability was associated with lower BMI throughout adulthood, with stronger associations found as participants aged within each cohort. Stronger associations were found in recent cohorts beyond mid-adulthood, but weaker associations were found at younger ages. Findings suggest that cognitive ability may be an increasingly important predictor of high BMI. Future studies should investigate the robustness of this association and its underlying mechanisms.