Data from a large British occupational cohort show that each of the 4 unhealthy behaviors examined—smoking, alcohol abstinence, low physical activity, poor dietary behavior—at any of the 3 measures over the 17-year follow-up was associated with poor executive function and memory in late midlife. Results suggest greater risk of poor executive function and memory with increase in the number of unhealthy behaviors. Furthermore, greater exposure from early to late midlife to individual or a combination of unhealthy behaviors was found to be associated with a greater risk of poor cognitive function in a dose-response manner.
Although there is some evidence showing that different health behaviors tend to cluster (34
), they have been studied for their combined impact on health only recently. Research suggests that the combined impact of unhealthy behaviors leads to a 3-fold to 4-fold increase in risk of mortality (29
), a 7-fold increase in risk of coronary heart disease in men (39
), and a 3-fold increase in risk of ischemic stroke in women (40
) compared with those with no unhealthy behaviors. Given these results, our findings are important because this is the first time that the association between a combination of health behaviors and cognitive function has been demonstrated. These results are in agreement with previous research showing greater risk of poor cognition in current smokers (2
), alcohol abstainers (4
), the physically inactive (10
), and those with poor dietary behavior (14
) when these have been examined separately. Our results suggest the importance of the combined impact of health behaviors for cognitive outcomes. In addition to the number of unhealthy behaviors, we examined their combinations, but there was no evidence to suggest that any particular combination of unhealthy behaviors drives the association. Thus, a simple summary score of unhealthy behaviors might be enough to assess association with cognitive outcomes.
Our results add to the previous evidence showing that health behaviors across the adult lifecourse may influence cognitive function at older ages (2
). These findings support the idea that midlife risk factors have a role in the development of cognitive impairment in later life (19
). Moreover, repeated measures of health behaviors allowed us to assess the association of cumulative exposure to unhealthy behaviors, separately and in combination, over midlife with cognition in late midlife. A greater risk of poor cognition was found with greater exposure to each of the health behaviors, as well as for combinations of health behaviors, suggesting that, not only the number of unhealthy behaviors but also the period of exposure should be taken into account in risk assessment.
It is increasingly clear that dementia has a long preclinical phase leading to increasing calls to examine the association between risk factors and cognitive outcomes many years before the clinical diagnosis of dementia (21
). Research suggests that individuals with mild cognitive impairment progress to clinically diagnosed dementia at an accelerated rate (25
), spurring research into cognition earlier in the lifecourse. In these studies, there is no clinical cutoff to define cognitive deficit, and 3 methods are commonly used to denote poor performance: either scores below 1.5 standard deviation from the mean (41
) or scores in the worst decile (42
) or worst quintile (43
). In the present study, we use the third definition with the objective of assessing the association with midlife poor cognition specific to the population studied.
In this study, 2 specific cognitive domains, executive function and memory, were found to be associated with health behaviors. Executive function, an umbrella term for various complex cognitive processes involved in achieving a particular goal (44
), has been shown to be particularly strongly affected in vascular dementia (45
). We assessed executive function using measures of reasoning and verbal fluency, as these tasks require the combination of different cognitive abilities such as memory, attention, and speed of information processing (31
). Health behaviors are important risk factors for vascular diseases (46
) and could influence executive function via the vascular pathway. In our data, there was also an association between health behaviors and memory, although it was less consistent. This difference could be explained by the fact that memory is likely to be less influenced by vascular risk factors (hypertension, cholesterol, diabetes) (47
). Another explanation could be better measurement precision for executive function as the score was made up of 3 tests compared with the single measure of memory.
The specific strengths of this study include a detailed prospective assessment of health behaviors over a 17-year period and adjustments for socioeconomic position, an important confounder that is associated with both health behaviors and cognition. However, at least 5 limitations for this study are noteworthy. First, although the sample covered a wide socioeconomic range, with annual full-time salaries ranging from £4,995 to £150,000 (1 British pound = 1.62900 US dollars), data are from white-collar civil servants and cannot be assumed to be representative of the general population. Second, during the 17-year follow-up, 50% of the baseline population was lost to follow-up, and baseline data suggest that these subjects were more likely to have unhealthy behaviors. Sensitivity analysis using phase 1 health behaviors suggests that the association between health behaviors and cognition could be underestimated in this study. Third, a U-shaped relation between alcohol consumption and cognition has been reported previously (50
). In our study, no higher risk was found in those who consumed more than 14 units per week. However, few participants of the Whitehall II study are heavy drinkers, and our data are best not used to pursue this further (4
). Fourth, the measure of physical activity changed between phase 1 and phase 5, which could explain the lack of association with physical activity at phase 1. Finally, the World Health Organization diet recommendation advises persons to eat at least 400 g of fruits and vegetables per day (51
), but we had only a measure of frequency of fruit and vegetable consumption and not the amount consumed.
In conclusion, our results show a prospective, cross-sectional, and cumulative association of smoking, alcohol abstinence, low physical activity level, and consumption of fruits and vegetables less than 2 times a day with poor cognitive function. Furthermore, we found that a greater number of these unhealthy behaviors were associated with a higher risk of poor cognition, particularly evident in executive function, and that this risk accumulated over the midlife. All these health behaviors are modifiable, and our results suggest that the promotion of a healthy lifestyle at all ages is important for cognitive outcomes.