Article Text


Smoking and cognitive change from age 11 to age 80
  1. I J Deary1,
  2. A Pattie1,
  3. M D Taylor1,
  4. M C Whiteman1,
  5. J M Starr2,
  6. L J Whalley3
  1. 1Psychology, School of Philosophy, Psychology and Language Sciences, University of Edinburgh, 7 George Square, Edinburgh EH8 9JZ, Scotland, UK
  2. 2Department of Geriatric Medicine, University of Edinburgh
  3. 3Department of Mental Health, University of Aberdeen, Aberdeen, Scotland, UK
  1. Correspondence to:
 Professor Deary;

Statistics from

Age related cognitive decline affects people’s quality of life and their ability to live independently.1 A recent review stated, “[we] are aware of no studies on the relationship between smoking and cognitive decline associated with normal aging or studies of the effect of smoking on cognition in normally aging individuals.”1 Some previous studies examined smoking in relation to pathological cognitive aging, but lacked cognitive data before the initiation of smoking, and used crude clinical cognitive assessments.2–4 Among middle aged subjects, current smoking was associated with poorer cognitive performance on tasks of psychomotor speed and cognitive flexibility.5 Smoking has been identified as a possible risk factor for accelerated cerebral degenerative changes, cognitive decline, and dementia.6 Here we show that smoking contributes to normal cognitive change from age 11 to age 80.

Participants, methods, and results

The Scottish Mental Survey of 1932 (SMS1932) tested mental ability in people born in 1921 (n = 87 498). The SMS1932’s Moray House test (MHT) was validated against the Stanford Binet test and includes verbal reasoning, numerical, spatial, and other items. From 1999 to 2001 we traced and retested 550 people from Edinburgh who were born in 1921 (the Lothian birth cohort 1921). All lived independently. We excluded people with mini-mental state examination scores below 24 and those with known dementia. We traced their scores on the MHT from SMS1932, readministered the MHT using the same instructions and time limit as the SMS1932, and collected information on smoking. In all, 470 people (194 men) provided full data.

We examined the effect of smoking on cognitive change from age 11 to age 80 using general linear modelling (analysis of covariance; SPSS version 11). Age corrected MHT score at age 80 was the dependent variable, smoking (never (n = 205); current (n = 34); ex-smoker (n = 231)) and sex were between subject variables, and age corrected MHT score at age 11 was a covariate. Among the current smokers the mean (SD) age at starting smoking was 18.9 (5.5) years (range 9 to 40). The ex-smokers’ mean age at starting smoking was 18.2 (5.2) years (range 7 to 60), and the mean age at stopping smoking was 49.6 (16.1) years (range 19 to 79 years). Only six of these ever-smokers (current and ex-) began smoking before the age of 11. The mean (SD) MHT scores for each smoking related subgroup at age 11 and age 80 are shown in table 1. MHT scores at age 11 had a large effect on scores at age 80 (F1,463 = 332.2, p < 0.001, η2 = 0.418). There was a significant, independent effect of smoking (F2,463 = 3.3, p = 0.039, η2 = 0.014), but not of sex (F1,463 = 3.1, p = 0.079, η2 = 0.007).2 The sex by smoking interaction was not significant (F2,463 = 1.7, p = 0.17, η2 = 0.007). Current smokers had significantly lower MHT scores at age 80 than never smokers (p = 0.013; mean difference = −5.2, 95% confidence interval (CI) −9.4 to −1.1) and ex-smokers (p = 0.016; mean difference = −5.0, 95% CI −9.0 to −0.9). These group comparisons remained similar in effect size and significance after entering years of full time education to the model.

Table 1

Moray House test scores at age 11 and age 80 by smoking status


Smoking affects cognitive change detrimentally from age 11 to age 80, with an effect that is similar in size to other contributors, such as the ε4 allele of the APOE gene.7 An advantage of this study is that the initial cognitive assessments were made when only a tiny percentage of the subjects had begun smoking. This finding adds to those of a previous study which found that, among middle aged participants, current smokers had reduced cognitive performance when compared with never smokers.5 In the present study, a history of having smoked and then given up smoking was not associated with any lowering of cognitive scores in old age. At age 80 there are survivor effects on cohorts owing to factors—such as death and illnesses—that are related to smoking. It might be expected that smokers in our cohort would be biased toward being especially fit and cognitively able. Thus selection bias could lead to our underestimating the effect of smoking on cognitive aging. The effect of smoking on cognitive aging might be direct, associated with, for example, biochemical factors such as antioxidant defences; neuropathological changes including acceleration of perfusional decline, cerebral atrophy, and polioaraiosis and leucoaraiosis (thinning of grey and white matter densities, respectively)6; or smoking related disease—though smoking did not explain the effect of cardiovascular disease on cognition in the Rotterdam study,4 nor unequivocally in the Zutphen study.3 It might also be indirect, being an indicator of a general tendency toward healthy lifestyle choices and responsiveness to health education. These possibilities notwithstanding, our data add to the reasons for giving up smoking, irrespective of age.


The Lothian birth cohort data were gathered with support from a Biotechnology and Biological Sciences Research Council Science of Ageing (SAGE) grant. IJD is the recipient of a Royal Society–Wolfson research merit award. LJW holds a Wellcome Trust career development award.

Competing interests: none declared

View Abstract

Request permissions

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.