Statistics from Altmetric.com
In their report van Oort and others1 clearly describe and empirically examine the potential mediating factors—broadly categorised by the authors as material, psychosocial, and behavioural—that might account for the well established inverse education-mortality gradient. We believe the role of intelligence (denoted here as IQ, and defined as a person’s ability to learn, reason, and solve problems2) warrants mention, given its link with all cause mortality, other somatic health outcomes, and at least two (material and behavioural factors) of the aforementioned pathways.3
Recently reported findings from a series of cohort studies show an inverse association between IQ, assessed using psychometric tests, and later death, whether this “exposure” was quantified in childhood, early adulthood, middle age, or older age.3 Although fewer data are available for cause specific outcomes, similar gradients have also been reported for childhood assessed IQ in relation to adult risk of ischaemic heart disease (but not stroke), selected cancers, and accidents.3 IQ has also been linked with behavioural factors, including smoking patterns, such that adults with higher early life IQ scores are more likely to subsequently give up the habit than their lower performing counterparts.4 While these findings are comparatively recent, the suggestion that early life IQ might influence later life material measures of socioeconomic position—particularly income,2 but also car and house ownership—has a long research tradition.
Based on these findings and using the authors’ own conceptual model (figure 1; page 2151), IQ may be regarded in at least three ways. Firstly, education may be a proxy for IQ. However, this is not to ignore potential interplay between IQ and education, such as mediation or moderation (effect modification), and the influence this might have on health. Secondly, in a related point, IQ might generate individual differences in educational attainment, in addition to being independently associated with material and behavioural factors. In studies that adjusted for education in the IQ-mortality relation, results are inconsistent with some investigators finding pronounced attenuation, while others do not.3 Thirdly, given that education may represent a cognitive archaeological “record” of pre-adult insults (for example, illness, nutritional privation, poor living conditions, psychosocial stress), it is probable that IQ, given its metric properties (education is normally quantified categorically), is a more sensitive marker of such exposures.5
In summary, there is new and persuasive evidence to link early cognitive ability and education with later health outcomes. Understanding the mechanisms that may underlie these associations should include an examination of whether education may be a partial mediator of, or a surrogate for, IQ differences.
We thank Marcus Richards for providing helpful comments on an earlier version of this correspondence.
We appreciate the comments of Batty and Deary on our paper reporting on the explanations of educational inequalities in mortality. We agree with them that there is evidence linking early cognitive ability to later health outcomes and that inclusion of education and early life intelligence may be important for studies of health inequalities. However, the latter seems to be dependent on the purpose of the study. It was the aim of our study to contribute to the debate on the importance of different mediating factors in the causal pathway of educational inequalities in health, and to provide more guidance to policy recommendations to reduce these inequalities, and it is less clear if inclusion of early life intelligence would have contributed much to these aims.
We believe that the most appropriate position for early life intelligence in our conceptual model would be preceding educational achievement. In this position early life intelligence could contribute to the understanding of differences in educational achievement. In contrast, its contribution to the understanding of causal pathways from education to health would be limited.
Inclusion of intelligence in a study like ours would increase understanding of the “core” roots of inequalities in health and should therefore be supported. Using the same arguments as the authors did however, this can be also be said from other factors, such as birth weight, parental socioeconomic position, and neighbourhood deprivation.1–3 These factors are also related to educational achievement, and do have mediating and interacting effects.
Finally, the connection of early life intelligence to policy recommendations is less straightforward than this is for education: recommending improvement of an individual trait as early life intelligence may be more difficult than recommending improvement of educational achievement, for example by policies that maximise the chance that children will remain at school.4
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.