Response to Ollson, Hulting, and Montgomery
We read with interest the article by Olsson et al. (1) that reported on the relation between IQ at age 11 years and risk of self-reported diabetes at age 42 years in the 1958 British birth cohort. Follow-up of large groups of children into adult life is important in this context because the measure of cognition can be regarded as being premorbid. This contrasts with assessment of cognition in older cohorts in which data interpretation is hampered by the possibility of reverse causality. As the authors indicate (1), although IQ in middle- and older-aged persons may influence diabetes, the reverse is also entirely feasible. The report by Olsson et al. (1) omitted the citations of a series of articles that have examined exactly the same and other highly relevant associations.
There was no relation between childhood IQ and self-reports of physician-diagnosed diabetes in the 1970 British birth cohort (2) (82 diabetes cases in 8,171 men and women) in which study members were followed up at age 30 years (odds ratio [OR] per one SD increase in IQ at age 10 years 0.89 [95% CI 0.72–1.11]) or in the Aberdeen Children of the 1950s Study (3) (89 diabetes cases in 5,340 men and women) in which participants were resurveyed at age 44–52 years (OR per one SD increase in IQ at age 11 years 0.93 [95% CI 0.73–1.18]). However, we also found evidence of an inverse relation of IQ at ∼20 years of age with both blood glucose (r = −0.092; P < 0.001) and the metabolic syndrome (OR per one SD increase in IQ at age 20 years 0.86 [95% CI 0.80–0.99]) in middle age (4).
In our own analyses of the 1958 British birth cohort, published in a book chapter in Brazilian Portuguese of which the authors could not have been aware (5), we also found an inverse IQ-diabetes association in analyses of women after controlling for early life factors such as parental occupational social class (OR highest IQ tertile versus lowest 0.37 [95% CI 0.15–0.92] P for trend = 0.001). When the study participants’ own socioeconomic position in adulthood (current social class and educational qualifications) was added to the multivariable model, this gradient was eliminated (OR 0.79 [95% CI 0.25–2.44] P for trend= 0.26). The same results were apparent in men. The pronounced reduction in the magnitude of the IQ–risk factor relation after adjusting for the adult social circumstances could implicate a pathway through which early life IQ may exert an influence on adult risk of chronic disease. It is also tenable that including education in our statistical model may be seen as overadjustment given that childhood mental ability and attained educational qualifications are moderately correlated (for instance, rs = 0.43, P < 0.001 in the 1970 British birth cohort ).
Although this body of work signals a growing interest in the link between early IQ and later health outcomes, it remains unclear whether preadult cognition has a role in the etiology of later diabetes.
- DIABETES CARE