Objective: To investigate possible mediators of the relationship between childhood
general cognitive ability scores and cognitive ability change and adulthood physical
Method: 5435 participants (2645 male, 2790 female) drawn from the National Child
Development Survey provided data on childhood cognitive ability aged eleven,
exercise frequency, smoker status, depression scores, Body Mass Index, general health
and occupational and educational level aged forty-two and self-report physical
function scores aged fifty. Mediation analyses tested the hypotheses that childhood
cognitive ability and lifetime cognitive ability change would significantly predict
being in the lowest sex-specific quintile of physical function scores and that the model
would differ by sex. Logistic regressions provided odds ratios.
Results: For men, childhood intelligence significantly predicted low physical function
scores via general health, SES, depression and smoking status. Odds ratios for the
lowest relative to highest general health rating was 8.42 (CI(95%) LB 4.27, UB
16.59). The highest SES group had an OR of 0.34 (CI(95%) LB 0.16, UB 0.74)
relative to the lowest group. For a single score increase in depression the OR was 1.13
(CI(95%) LB 1.09, UB 1.20) and for exercise frequency 0.932 (CI(95%) LB 0.89, UB
0.98). When CA change was the predictor, depression was no longer a significant
mediator (p=.17). The direct effect of intelligence disappeared after accounting for the
indirect effects (childhood CA: p=.469, CA change: p=.82).
In women, childhood CA significantly predicted midlife PF via all mediators except
exercise frequency and SES. This was the same when CA change was the predictor
except smoker status became non-significant (p=.105). Odds ratios for 1SD increase
in childhood CA was 0.85 (CI(95%) LB 0.76, UB 0.96), for BMI score 1.10 (CI(95%)
LB 1.08, UB 1.12), and for depression score 1.10 (CI(95%) LB 1.07, UB 1.13).
Current smokers were 1.40 times as likely as non-smokers (CI(95%) LB 1.09, 1.80).
The lowest general health group had an odds ratio of 11.41 (CI(95%) LB 5.98, UB
21.77). The highest educated cut their risk to 0.34(CI(95%) LB .14, UB 0.86). The
direct effects remained even after accounting for the indirect effects (childhood CA:
β=-.183, p=.0024, CI(95%) LB -.301, UB -.065, 53.63% explained; CA change: β=-.395, p= , CI(95%) LB , UB, 33% explained).
The results suggest that general intelligence and its decline are significant risk factors
for developing low physical function in midlife via both health-related and