Early-life cumulative exposure to excess bodyweight and midlife cognitive function: longitudinal analysis in three British birth cohorts

Abstract

Background: Excess bodyweight (BMI >25 kg/m²) in midlife (age 40-65 years) has been linked to future cognitive decline and an increased risk of dementia. Whether chronic exposure to excess bodyweight in the early decades of life (<40 years) is associated with compromised cognitive function by midlife, however, remains unclear. This study therefore aimed to test potential bidirectional direct and indirect pathways linking cumulative exposure to excess bodyweight and cognitive function in the early decades of life.

Methods: In this longitudinal analysis, harmonised measures of BMI and cognitive function were available in 19 742 participants aged 47-53 years recruited to the 1946 National Survey of Health and Development (n=2131), the 1958 National Child Development Study (n=9385), and the 1970 British Cohort Study (n=8226). Individual BMI trajectories spanning three decades from age 10-40 years were created for each participant and excess bodyweight duration, BMI change between ages, and cumulative excess bodyweight exposure were calculated. Harmonised measures of verbal and non-verbal ability, mathematical ability, and reading ability were used to create a latent factor for childhood cognitive function, and immediate and delayed recall, animal naming, and letter-search speed tests were used for midlife cognitive function. Multivariable linear regression and structural equation models (SEM) were used to test for potential bidirectional relationships between cognition and excess bodyweight in both individual cohorts and pooled datasets while accounting for other potential early-life confounders.

Findings: Increases in BMI during adolescence and greater cumulative exposure to excess bodyweight across early life were associated with lower midlife cognitive function in all cohorts (eg, pooled difference in cognitive function per 10 years excess bodyweight duration -0·10; 95% CI -0·12 to -0·08; p<0·001). Further adjustment for childhood cognitive function attenuated many of these associations towards the null (eg, pooled difference in cognitive function per 10 years excess bodyweight duration -0·04; 95% CI -0·06 to -0·02; p=0·001), however, with any remaining associations then fully attenuating once further adjusted for other early-life factors (eg, pooled difference in cognitive function per 10 years excess bodyweight duration 0, -0·03 to 0·01; p=0·38). In the reverse direction, low childhood cognition was associated with greater cumulative exposure to excess bodyweight over the next four decades, although much of this relationship was found to probably be explained via other potentially modifiable upstream early-life factors such as childhood disadvantage. SEM in all cohorts suggested the presence of modest direct and indirect pathways connecting earlier cognitive function to later excess bodyweight, but scarce evidence for an effect of early-life excess bodyweight on cognitive function by midlife.

Interpretation: The association between cumulative exposure to excess bodyweight in early life and lower cognitive function in midlife is probably confounded by a persistently lower cognitive function from childhood. Initiatives to improve early-life factors such as childhood disadvantage and education, however, might exert dual but independent benefits on both of these factors before old age.

Funding: Alzheimer's Research UK, Diabetes Research and Wellness Foundation, Diabetes UK, British Heart Foundation, and Medical Research Council.

Figure 1

Study timeline and data collection (A) Timeline of study visits in all three cohorts from conception to follow-up, with each dot representing a point at which BMI was either directly recorded or self-reported (top panel). (B) Example of individual child–adult BMI trajectory from which duration of and cumulative exposure to excess bodyweight can be calculated. Delta signifies change in BMI between respective ages. AUC=area under the curve for cumulative exposure to excess bodyweight. BCS70=1970 British Cohort Study. NCDS=1958 National Child Development Study. NSHD=1946 Medical Research Council National Survey of Health and Development.

Figure 2

Early-life BMI trajectories categorised by tertiles of childhood cognitive function Simple line graphs connecting mean predicted values of BMI z-score at various ages across early life based on tertile of childhood cognitive function at age 10 years. Highest tertile of childhood cognitive function is indicated by blue lines, intermediate tertile is indicated by green lines, and lowest tertile is indicated by red lines. (A) NSHD, NCDS, and BCS70 adjusted for sex only. (B) NSHD, NCDS, and BCS70 adjusted for parents’ BMI, child's birthweight, childhood socioeconomic status, and childhood household overcrowding. (C) Pooled data from all three cohorts adjusted for sex. (D) Pooled data from all three cohorts adjusted for sex, parents’ BMI, child's birthweight, childhood socioeconomic status, and childhood household overcrowding. NSHD=1946 Medical Research Council National Survey of Health and Development. NCDS=1958 National Child Development Study. BCS70=1970 British Cohort Study.

Figure 3

Cross-lagged structural equation models to assess potential bidirectional relationships between excess bodyweight and cognitive function during early life All variables are adjusted for sex, childhood household overcrowding, and childhood socioeconomic status. Data are presented as standardised estimates. Maximum likelihood with missing values were used to account for missing data. RMSEA was 0·06 for NSHD, 0·06 for NCDS, and 0·05 for BCS70. CFI was 0·95 for NSHD, 0·95 for NCDS, and 0·96 for BCS70. TLI was 0·91 for NSHD, 0·92 for NCDS, and 0·93 for BCS70. The dashed grey line indicates non-significant pathways; the black line indicates p<0·05. Childhood cognition was expressed as a latent variable derived from harmonised cognitive measures consisting of VR, NVR, Mth, and RD at ages 10–11 years. Midlife cognition was expressed as latent variable derived from harmonised measures of AN, LS, Mi, and Md at ages 47–53 years. NSHD=1946 Medical Research Council National Survey of Health and Development. AN=animal naming. BCS70=1970 British Cohort Study. CFI=comparative fit index. LS=letter search speed. Md=delayed recall. Mi=immediate recall. Mth=mathematical ability. NCDS=1958 National Child Development Study. NVR=non-verbal reasoning. RD=reading ability. RMSEA=root mean square error of approximation. TLI=Tucker-Lewis Index. VR=verbal reasoning.