Pre-morbid intelligence, the metabolic syndrome and mortality: the Vietnam Experience Study
We examined the relationship between pre-morbid intelligence quotient (IQ) and the metabolic syndrome, and assessed the role of the metabolic syndrome as a mediating factor in the association of IQ with total and cardiovascular disease (CVD) mortality.
In this cohort study, 4,157 men with IQ test results from late adolescence or early adulthood [mean age (range) 20.4 (16–30) years] attended a clinical examination in middle-age [38.3 (31–46) years] at which the components of the metabolic syndrome were measured. They were then followed for 15 years to assess mortality.
In age-adjusted analyses, IQ was significantly inversely related to four of the five individual components comprising the metabolic syndrome: hypertension, high BMI, high triglycerides and high blood glucose, but not low HDL-cholesterol. After controlling for a range of covariates that included socioeconomic position, higher IQ scores were associated with a reduced prevalence of the metabolic syndrome itself (odds ratio1 SD increase in IQ 0.87, 95% CI 0.78–0.98). Structural equation modelling revealed that education was not a mediator of the relationship between IQ and the metabolic syndrome. The metabolic syndrome partially mediated the relationship between IQ and CVD but not that between IQ and total mortality.
In this cohort, higher scores on a pre-morbid IQ test were associated with a lower prevalence of the metabolic syndrome and most of its components. The metabolic syndrome was a mediating variable in the IQ–CVD relationship.
KeywordsCardiovascular disease CVD Intelligence quotient IQ Metabolic syndrome Mortality Socioeconomic position
principal components analysis
Empirical research in the field of cognition (i.e. intelligence quotient, IQ), has a long tradition in the social sciences. It has recently been suggested that the skills captured by IQ tests, such as comprehension and reasoning, may have a role in an individual’s interpretation of health promotion advice and hence their choice of behaviours . Thus, in comparison to their lower-performing counterparts, higher IQ-scoring individuals have more favourable levels of cigarette smoking [2, 3, 4, 5] (uptake and cessation), physical activity  and dietary characteristics [6, 7]. Although there are fewer studies exploring the link between IQ and physiological variables, it has been suggested that higher IQ scores are associated with lower levels of blood pressure or hypertension [3, 8, 9] and obesity [10, 11, 12]. The scant reports of the relationship between IQ and later measurement of serum cholesterol and high blood glucose or diabetes  reveal null results, although sample sizes are modest. While some of these risk indices represent the cluster of factors comprising the metabolic syndrome, to our knowledge, the relationship between IQ and this constellation of risk indices has not been examined.
Data from the Vietnam Experience Study, a large cohort of former US army personnel who had their IQ assessed at entry to the service in early adulthood and then took part in a telephone interview and medical examination some 20 years later [14, 15, 16], provide a novel opportunity to examine the IQ–metabolic syndrome relationship. As the vital status of these men post-medical examination is also known, an additional novel objective of the present analyses is to examine the mediating role, if any, of the metabolic syndrome in the association between IQ and both total and cardiovascular disease (CVD) mortality.
Data collection in late adolescence or early adulthood
Study participants were identified retrospectively using existing military records, a process that has been described in detail elsewhere [14, 15, 16]. In brief, information pertaining to place of service, military rank, ethnicity and IQ were extracted from military archives for 18,313 former military personnel.
On enlistment into military service, the men routinely completed a general aptitude (IQ) test: the Army General Technical Test. This consists of two subtests, verbal and arithmetic reasoning. We validated this test by comparing scores from it with those from components of the Wechsler Adult Intelligence Scale, a comprehensive and widely used test of cognitive ability that study participants (n = 4,411) completed during the medical examination in middle-age (described below). Total Army General Technical Test results were strongly correlated with information subtest scores (r = 0.74, 95% CI 0.73–0.76), block design scores (r = 0.51, 95% CI 0.49–0.55) and overall results (r = 0.74, 95% CI 0.73–0.76) from the Wechsler Adult Intelligence Scale. Mean age at army entry when IQ was assessed was 20.4 years (range 16–30).
Data collection in middle-age—telephone survey
A total of 15,288 (85.6% of target population) of 17,867 men found to be alive on 31 December 1983 participated in the 1985 telephone survey, during which enquiries were made about the study participants’ medical history, health behaviours and socioeconomic characteristics. Socioeconomic position was measured using household income, an index of occupational prestige [17, 18] and years of completed education.
Data collection in middle-age—medical examination
In 1986, a random sample of telephone interview respondents (n = 6,443) was invited to attend a 3 day medical examination; 4,462 attended (69.3% of those invited; mean age 38.3 years, range 31–46). All men were requested to fast from 19:00 hours on the evening before medical testing. Blood was drawn the following morning and levels of triglycerides and cholesterol fractions ascertained using an autoanalyser (Ektachem 700; Kodak, Rochester, NY, USA) [19, 20]. Serum glucose level was determined with a standard adaptation of the glucose oxidase–peroxidase–chromogen-coupled system for glucose determination in biological fluids [19, 20]. Blood pressure, height and weight were assessed using standard protocols. Body mass index was computed using the usual formula (weight [kg]/ height [m]2). Subsequent to completion of the medical examination, attempts were made to match study participants against mortality databases for a 15 year period.
Definition of the metabolic syndrome
We used two methods to define the metabolic syndrome. In the first method, we defined the metabolic syndrome and its components using a modified version of the Adult Treatment Panel III recommended diagnostic criteria  (here, termed the conventionally derived index). According to this definition, participants were classified as having the metabolic syndrome if any three of the following were present: BMI > 30 kg/m2 (in the absence of data on waist circumference, BMI at this threshold is regarded by the WHO as an acceptable substitute in defining the metabolic syndrome ); fasting plasma glucose ≥6.1 mmol/l (110 mg/dl) or medication for diabetes (as reported at the medical examination); triglycerides ≥1.7 mmol/l (150 mg/dl); HDL-cholesterol level <1.036 mmol/l (40 mg/dl); and blood pressure ≥130/85 mmHg and/or use of antihypertensive medication. A second approach was based on evidence that a single latent factor may underlie the core components of the metabolic syndrome . We therefore carried out a principal components analysis (PCA) of BMI, triglycerides, HDL, blood glucose, systolic and diastolic blood pressure, and use of antihypertensive medication. To ascertain whether a single underlying component was sufficient to describe the associations among these variables, we used the scree slope criterion . The scree slope showed just one substantial component above the ‘scree’ of trivial components and we therefore extracted a single score based on the first unrotated principal component (here, termed the PCA-derived index).
Comparison of early adult characteristics of men included in the analytical sample with those excluded
Included (n = 4,157)
Excluded (n = 11,131)
Army income (US$ per week)
Place of service
Ever served in Vietnam
Other overseas posting
Served in USA only
IQ at enlistment, mean (SD)
Age at enlistment (years), mean (SD)
We assessed the relationship between IQ and the metabolic syndrome in three ways that provided complementary information. In the first analyses, we used logistic regression to examine the relationship between IQ and both each individual component of the metabolic syndrome and the conventionally derived metabolic syndrome. In these analyses, outcomes were dichotomised and adjustments were made for a range of covariates (age, rank, ethnicity, education, income and social prestige). In the second analyses, by computing Pearson partial correlation coefficients, we related IQ to the naturally coded (continuous) versions of the components of the metabolic syndrome and the PCA-derived index described above, which represents a score for the metabolic syndrome. This preserves all the information in the variables and provides effect sizes for associations. Third, we carried out an analysis using structural equation modelling with the EQS program, version 6.1 . In the hypothetical model, a latent trait of general cognitive ability was formed from the verbal and arithmetic ability test scores. The following indicators were used to form a latent trait of metabolic syndrome: systolic blood pressure, BMI, triglycerides, HDL-cholesterol, glucose and hypertensive medications. The explicit hypothesis tested was that latent general mental ability influenced the latent metabolic syndrome trait and that some of that effect was mediated via education, a close correlate of IQ . The measurement models (for the latent traits) and the structural model (the path part of the model) are given in Fig. 1. The fit of the model was tested comprehensively, as described in the results section.
Finally, to examine the mediating role of the metabolic syndrome, if any, between IQ and mortality from all causes and CVD we computed hazard ratios (HRs) and accompanying 95% CIs using Cox proportional hazards regression model with age as the underlying time scale . IQ was used as a continuous measure with effect estimates reported for a 1 SD increase. Follow-up time was taken from the medical examination until censoring, death or 31 December 2000, whichever came first (mean follow-up time 15.1 years).
The Vietnam Experience Study was authorised by the US Congress. The protocol for the study was reviewed by the US Office for Technology Assessment, the Department of Health and Human Sciences Advisory Committee, the Agent Orange Working Group Science Panel and a panel from the US Centers for Disease Control. On the first day of the medical examination, participants attended an orientation session and signed a consent form.
IQ and study characteristics in the Vietnam Experience Study (n = 4,157)
IQ scoreb, mean (SD)
Data recorded at baseline
Age group (years)
Army income (US$ per week)
Place of service
Ever served in Vietnam
Other overseas posting
Served in USA only
Data recorded at follow-up
Grade ≤ 11
Grade ≥ 13
Family income (US$ per year)
ORs (95% CI) for the relationship of a 1 SD increase in IQ with components of the metabolic syndrome and the conventionally derived metabolic syndrome in the Vietnam Experience Study (n = 4,157)
High blood glucose
With condition, n (%)
Age- + rank-adjusted
Age- + ethnicity-adjusted
Age- + place of service-adjusted
Age- + education-adjusted
Age- + social prestige-adjusted
Age- + income-adjusted
Age- + socioeconomic positiona
Age- + fully adjusted
HRs (95% CI) for the relationship of a 1 SD increase in IQ with total and cardiovascular disease mortality in the Vietnam Experience Study
HR (95% CI)
Per cent reduction in HRd
HR (95% CI)
Per cent reduction in HRd
Age- + ethnicity-, place of service-, army income-adjusted
Age- + systolic blood pressure-adjusted
Age- + BMI-adjusted
Age- + blood glucose-adjusted
Age- + HDL-cholesterol-adjusted
Age- + triglycerides-adjusted
Age- + adult socioeconomic positionc-adjusted
Age- + conventionally derived-adjusted
Age- + PCA-derived-adjusted
To our knowledge, this is the first study to examine the link between IQ and the metabolic syndrome, and to assess the impact, if any, of controlling for the metabolic syndrome on the IQ–mortality gradient. We found that men with higher IQ test scores in early adulthood experienced a reduced prevalence of the metabolic syndrome in middle-age. This gradient was not explained by control for markers of socioeconomic circumstances either at the time of IQ testing (army income) or during measurement of the components of the metabolic syndrome (income, occupation, education). We did not find evidence that the IQ–metabolic syndrome association is mediated by education. Controlling for the metabolic syndrome had little impact on the IQ–total mortality relationship, with greater attenuation seen for IQ–CVD. This might be ascribed to CVD being more strongly linked with the metabolic syndrome than all-cause mortality , the latter comprising a range of outcomes that will dilute the strength of the association.
As described, early life IQ is associated with more favourable levels of smoking [2, 3, 4, 5], physical exertion , dietary characteristics [6, 7] and heavy alcohol consumption , although the latter is not a universal finding [32, 33, 34]. This suggests that the skills captured by IQ tests, such as verbal comprehension and reasoning, may be important in the successful management of a person’s health behaviours. As such, it is perhaps unsurprising that IQ is also associated with several physiological correlates of these behaviours, e.g. blood pressure, adiposity, blood glucose and cholesterol, which collectively represent the metabolic syndrome.
Study strengths and limitations
This study has a number of strengths, not least its prospective design and the temporal measurement of IQ, which preceded that of the metabolic syndrome and, in turn, mortality. Moreover, we used an appropriate combination of epidemiological and structural equation models to fully understand the associations and possible confounding and mediating factors. Having IQ test scores from late adolescence or early adulthood, rather than from older ages, is also crucial in examining the link between cognitive ability and the metabolic syndrome. Where mental test data are available from earlier in life, any apparent protective effect of high IQ scores is unlikely to be explained by reverse causality, in which components of the metabolic syndrome themselves could actually lead to a reduction in cognitive function .
The study is limited in that women were not examined. It is uncertain to what extent these results might be applicable to women. Second, the range of IQ scores in the present cohort may be narrower than in general population groups, not only due to the occupational nature of the population studied (the so called ‘healthy worker’ effect), but also because, from the outset, personnel with a rank corresponding to above sergeant were excluded from the study sample. However, any reduced variance would lead to an underestimation of the true effect of IQ on the metabolic syndrome. Third, to our knowledge, the original IQ tests were in the English language, which may have handicapped ethnic minorities.
In conclusion, in this first study to examine the link between IQ and the metabolic syndrome, we found an inverse relationship, which was not mediated by education. The metabolic syndrome did not, however, appear to substantially mediate the relationship between high IQ scores and lower mortality rates, although some attenuation was apparent when CVD death was the endpoint of interest.
G. D. Batty is a Wellcome Trust Fellow; M. Shipley is supported by the British Heart Foundation; I. J. Deary is the recipient of a Royal Society-Wolfson Research Merit Award.
Duality of interest
The authors declare that there is no duality of interest associated with this manuscript.
- 4.Batty GD, Gale CR, Chandola T, Deary IJ (2007) Early life IQ, risk factors, and later cardiovascular disease: new data from the 1958 British Birth Cohort and a systematic review. In: Alves JG, Sampaio-Carneiro M (eds) Prevention of adult chronic diseases in infancy and adolescence (in Brazilian Portuguese). Medbook, Rio de Janeiro, pp 29–46Google Scholar
- 14.The Centers for Disease Control Vietnam Experience Study (2004) Postservice mortality among Vietnam veterans. JAMA 257:790–795Google Scholar
- 16.Batty GD, Shipley MJ, Mortensen L et al (2008) IQ in late adolescence/early adulthood, risk factors in middle-age, and later all-cause mortality in men: the Vietnam Experience Study. J Epidemiol Community Health (in press)Google Scholar
- 17.Duncan O (1961) A socioeconomic index for all occupations. Free Press, New YorkGoogle Scholar
- 21.Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (2001) Executive summary of the third report of The National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). JAMA 285:2486–2497CrossRefGoogle Scholar
- 24.Kline P (1994) An easy guide to factor analysis. Routledge, LondonGoogle Scholar
- 25.Bentler P (1995) EQS structural equations program manual. Multivariate Software, EncinoGoogle Scholar
- 27.Cox DR (1972) Regression models and life-tables. J R Stat Soc [Ser B] 34:187–220Google Scholar
- 28.Kline R (2005) Principles and practice of structural equation modeling. Guilford, New YorkGoogle Scholar
- 34.Batty GD, Deary IJ, Schoon I et al (2008) Childhood mental ability in relation to alcohol drinking problems and consumption in adulthood: the 1970 British Cohort Study. Am J Public Health (in press)Google Scholar