Psychosomatic Medicine
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Published online before print September 10, 2007, 10.1097/PSY.0b013e31814c3e7c
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrowRequest Permissions
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Shipley, B. A.
Right arrow Articles by Deary, I. J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Shipley, B. A.
Right arrow Articles by Deary, I. J.
Related Collections
Right arrow Other Epidemiology
Right arrow Cognitive Functioning
Psychosomatic Medicine 69:640-650 (2007)
© 2007 American Psychosomatic Society

ORIGINAL ARTICLES

Association Between Mortality and Cognitive Change Over 7 Years in a Large Representative Sample of UK Residents

Beverly A. Shipley, MA, MPhil, Geoff Der, MA, MSc, Michelle D. Taylor, MSc, PhD and Ian J. Deary, PhD, FRCPE

From the Psychology (B.A.S, M.D.T, I.J.D), School of Philosophy, Psychology and Language Sciences, University of Edinburgh, Edinburgh, Scotland; MRC Social and Public Health Sciences Unit (G.D.), University of Glasgow, Glasgow, Scotland.

Address correspondence and reprint requests to Ian J. Deary, Psychology, School of Philosophy, Psychology and Language Sciences, University of Edinburgh, 7, George Square, Edinburgh EH8 9JZ, UK. E-mail: ian.deary{at}ed.ac.uk


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 RESULTS
 CONCLUSIONS
 NOTES
 REFERENCES
 
Objective: To examine the association between change in reaction time and cognitive performance over 7 years and the risk of death from all causes and some specific causes after controlling for known risk factors.

Methods: The sample comprised members of the Health and Lifestyle Survey (HALS) of community-dwelling adults in England, Scotland, and Wales. Baseline testing (HALS1), involving 9003 people, took place in 1985 and 1986. Sociodemographic, lifestyle, health, and physiological information was collected. Cognitive functioning was measured using tests of simple and choice reaction time, a short memory test, and a test of visual-spatial reasoning. Follow-up testing (HALS2) took place in 1991 and 1992, when 5352 members of the study were administered the same questionnaires, physiological examinations, and cognitive tests. The sample has been followed for mortality up to June 2005.

Results: After controlling for age, gender, and the relevant baseline cognitive test scores, greater declines between HALS1 and HALS2 on simple reaction time mean and variability, choice reaction time mean and variability, memory and visual-spatial reasoning were associated with significantly increased risks of death from all causes, all cardiovascular diseases (CVDs), coronary heart disease (CHD), stroke, and respiratory disease. These associations were only slightly attenuated after adjusting for occupational social class, educational, smoking, alcohol consumption, physical activity, body mass index, blood pressure, and lung function.

Conclusions: Decline in performance of reaction times and simple cognitive tasks across a 7-year period was associated with an increased risk of death from all causes, all CVDs, CHD, stroke, and respiratory disease up to 13 years later, even after adjustment for known risk factors.

Key Words: reaction time • intraindividual variability • cognitive change • mortality • information-processing speed • social class

Abbreviations: RT = reaction time; HR = hazard ratio; CI = 95% confidence interval; HALS = Health and Lifestyle Survey; CVD = cardiovascular disease; CHD = coronary heart disease; BMI = body mass index; BP = blood pressure; FEV = forced expiratory volume; SRT = simple reaction time; CRT = choice reaction time.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
RESULTS
 CONCLUSIONS
 NOTES
 REFERENCES
 
Lower cognitive scores are related to increased rates of all-cause and cause-specific mortality (1–5). However, the evidence generally comes from studies where cognition was measured cross-sectionally and it is possible therefore that the relation is due to confounding with preexisting morbidity. Currently, little is known about the relationship between cognitive change and health, particularly in adults <65 years, as the literature is generally confined to older samples and focuses on all-cause mortality.

An association between cognitive decline and mortality was first described by Kleemeier (6); Riegel and Riegel (7) first proposed the "terminal drop hypothesis." The hypothesis proposes that cognition begins to decline about 5 years before death and this "critical" decline in cognition is predictive of mortality (8). However, Palmore and Cleveland (9) proposed that there is no terminal drop in cognition before death; rather, cognition declines continuously for several years before death.

A recent review on cognitive decline and mortality (3) found nine longitudinal studies (5,8,10–17) based on general population samples. Johansson and Berg (11) found decreases in memory over a period of 9 years predicted subsequent mortality in 120 adults aged around 70 years. Significant associations between cognitive decline and mortality were also found by Deeg and associates (8). Memory change over 8 years was found to be an independent predictor of mortality in 211 individuals aged ≥70 years. Conversely, in a larger study of 1210 participants of the Seattle Longitudinal Study, Bosworth et al. (16) found no association between change in cognition and mortality. Similar findings were also noted by Hassing et al. (18), who found no association between cognitive change and mortality over 6 years. Since Bosworth and Siegler's (3) review, Wilson et al. (19) examined the relationship between cognitive decline and mortality in 763 clergy members (age range = 70–85 years). After controlling for baseline health, they found a significant association between greater cognitive decline and increased mortality during 5 years of follow-up.

Furthermore, a recent issue of the European Psychologist (20) published six papers (21–26) under the title "Death and Cognition," of which two papers examined the association between cognitive change and mortality. Using 20-year follow-up data from the Manchester Longitudinal Study, Rabbitt et al. (21) examined the association between change in the AH4 and mortality in 5842 individuals aged 49 to 93 years. Average performance declines of 10% and 20% between successive test sessions were associated with two times and two and a half times the risk of dying, respectively. Ghisletta et al. (25) also found that the level of cognitive change predicted survival. Greater decline on tests of perceptual speed and verbal fluency was associated with increased mortality after controlling for age, gender, and socioeconomic status in 516 adults >70 years old.

One measure of cognition rarely used in studies on cognition and mortality is reaction time (RT), which is thought to examine some aspect of brain information processing (27). Compared with psychometric mental tests, with which they have a moderately strong correlation (27), RTs involve responses to simple stimuli and are less obviously confounded by cultural and social background. RTs become slower and more variable with age (28,29), and it is possible that age-related differences in RTs may account for changes in more complex cognitive functions (30). In two previous reports using data from the Health and Lifestyle Survey (HALS), we found that slower and more variable RTs at baseline were significantly associated with increased all-cause and cause-specific mortality (1,2). In the Scottish Twenty-07 study (27), slower RTs and lower performance on the AH4 (31) were also significant predictors of increased mortality in middle-aged adults. Furthermore, after controlling for RT, the association between IQ and mortality was reduced to nonsignificance.

The evidence suggests that cognition, and possibly also change in cognition, are associated with risk of mortality. RTs are associated with risk of mortality and account for some of the cognition-mortality risk. What remains uncertain is the association between RT change and mortality. Does change in this simple index of information processing efficiency offer a signal about the more general health of the organism? The HALS recorded RT twice, 7 years apart. Therefore, it offers the rare opportunity to examine whether change in RT, adjusted for baseline performance, is associated with risk of death. HALS is also a large population study that includes adults of all ages. Previous literature on cognitive decline and mortality has concentrated on older samples (5,8,10,11,13, 16,25,32,33). The HALS has a long follow-up period for mortality (13 years) compared with previous studies in the area (8,11,13,16,32,34). The mortality data in the HALS provide information on specific causes of death, which might suggest mechanisms for the cognition-mortality association. Therefore, the present study examines cognitive, especially RT, changes over a 7-year period in 3802 individuals of all adult ages and relates these changes to risks of all-cause and cause-specific mortality over 13 years after controlling for baseline cognitive function.

Methodology
Participants and Procedure
The UK HALS, a survey of adults in England, Scotland, and Wales, was initially conducted in 1985 and 1986 with a follow-up 7 years later in 1991 and 1992. At baseline (HALS1), 12,254 addresses were chosen from UK electoral registers and one adult from each household was invited to participate in the study. Initial interviews were conducted on 9003 individuals. The study sample was compared with the 1981 Census to determine whether it accurately represented the UK population. When compared with the UK general population, the study population (n = 9003) included slightly more women, fewer single and divorced individuals, and fewer individuals in the lowest social class and with the lowest educational attainment. However, these biases were small; in general, the sample was a reliable representation of the UK population (35).

The first study interview of HALS1 was completed in participant's homes where information was collected on sociodemographic factors and health behaviors. A second home visit involved a physiological examination and cognitive testing. A total of 7414 individuals had complete physical and RT data at HALS1. The same procedures were repeated at HALS2. Attrition due to factors such as death, moving abroad, refusals, illness, or loss-to-follow-up, resulted in 5352 individuals completing both waves (36). Those who completed testing at HALS2 were more likely to be of higher social class, younger, and healthier. When compared with the 1991 Census, the HALS2 study population included greater numbers of middle-aged individuals and fewer younger and older individuals (37). Of the 5352 individuals who had both HALS1 and HALS2 data, 3802 had complete cognitive, physical, and sociodemographic data at both time points. This subsample was comparable with the 5352 individuals who completed HALS2. Full details of HALS1 (35) and HALS2 (37) can be found elsewhere.

All participants were continuously followed for mortality. The latest available follow-up was completed in June 2005, giving 13 years of mortality follow-up since the completion of HALS2.

Measures

Cognitive Factors
Reaction Time
A portable, battery-operated device was used to measure simple and four-choice RTs. The device consisted of an LCD screen and five buttons numbered 1, 2, 0, 3, and 4 from left to right (1). Simple RT (SRT) was the time taken to press the "0" key after being presented with a "0" stimulus. Choice RT (CRT) was the time taken to press one of four buttons (1–4) after being presented with one of the four digits on the screen. In both cases, participants completed eight practice trials. This was followed by 20 test trials for SRT and 40 test trials for CRT. Inter-stimulus interval for both SRT and CRT varied between 1 and 3 seconds (1).

All SRT and CRT mean and standard deviations (SD) (intra-individual variabilities) were measured in milliseconds and standardized to a mean of zero and unit SD.

Verbal Declarative Memory
During a discussion about the fiber content of food, a list of ten common foods was read out (roast meat, Digestive biscuits, potatoes, eggs, orange juice, grilled fish, Weetabix, white bread, cheese, apples). After a period of a few minutes, participants were asked to recall the list, which was recorded.

Visual-Spatial Reasoning
Visual-spatial reasoning was measured using a block counting test that consisted of six three-dimensional drawings of piles of blocks (1). Participants had to "count" the number of blocks that made up each pile.

Sociodemographic Factors
All sociodemographic information was taken from HALS2.

Age was recorded in years.

Occupational social class was based on the occupation of the head of the household or the usual occupation of the retired or unemployed head of household. It was classified using the Registrar General's occupational social class and comprised of six categories: professional (I); managerial (II); skilled nonmanual (IIIN); skilled manual (IIIM); semi-skilled manual (IV); and unskilled (V).

Level of education was based on the highest qualification obtained by the participant either while at school or since school. It was classified into five levels: first or higher degree; semiprofessional or professional qualification, for example, nursing or teaching qualifications; A-level or equivalent, including advanced City and Guilds certificates; O-level or equivalent, including ordinary City and Guilds certificates; no educational qualifications (work-related certificates were included in this category).

Health Behavior Variables
Current smoking status was categorized as: never smoked; ex-smoker; current occasional smoker (<1 cigarette per day); and current regular smoker (≥1 cigarette per day).

In the week before the interview, total alcohol consumption was measured in alcohol units and by type of drinker. For total alcohol consumption, 1 unit (8.5 g) of pure alcohol was measured as one half pint of beer or cider, a single measure (25 ml) of spirits, a glass (125 ml) of wine, or a small glass (50 ml) of fortified wine. Type of drinker was categorized as: nondrinker; very special occasions drinker; occasional drinker; and regular drinker.

Participation in physical activity was recorded as minutes spent doing vigorous activity; occasions spent doing vigorous activity; minutes spent doing nonvigorous activity; and occasions spent doing nonvigorous activity for a range of activities including work-associated physical activity, week day and weekend walking, housework, gardening, home improvement, and various sports activities.

Health Status Variables
Body mass index (kg/m2) (BMI) was calculated from height, measured in meters, using a portable stadiometer, and weight in kilograms using electronic scales.

Median systolic and diastolic blood pressure (BP) values were computed from four recordings of BP taken at 1-minute intervals using an automatic BP monitor (Accutorr, Datascope Corp., Montvale, New Jersey).

Forced expiratory volume (FEV) in 1 second in liters was measured using a portable electronic spirometer. To allow for individual differences in body size, this value was divided by the square of height in meters.

Vital Status
Members of HALS are flagged by the National Health Service central registry, which also provides copies of death certificates. Although death certificates provided up to three primary and two secondary causes of death, only the "underlying cause" of death was used. This was classified using the Ninth International Classification of Diseases (ICD9) and was categorized as all cardiovascular diseases (CVD) (ICD9 codes 390–459), coronary heart disease (CHD) (ICD9 codes 410–414, i.e., a subcategory of CVD), stroke (ICD9 codes 430–438, i.e., a subcategory of CVD), respiratory disease (ICD9 codes 460–519), lung cancer (ICD9 code 162), and all nonlung cancers (ICD9 codes 40–161 and 163–208).

Statistical Analysis
We used regression to obtain measures of RT and other cognitive change that were unrelated to baseline scores (38). For each RT and other cognitive score, the HALS2 measure was the dependent variable in a linear regression model and the HALS1 measure was the independent variable. The standardized residuals from these models were utilized as measures of cognitive change between HALS1 and HALS2. They were then used in Cox's regression models, which also included HALS1 cognitive function and other covariates, to calculate hazard ratios for the proportionate change in all-cause and cause-specific mortality risk by level of change in the RT parameters or cognitive function measures. The hazard ratio units from the Cox's regression models are SD values of the residuals from the regression of HALS2 cognitive values on HALS1 cognitive values. Therefore, the hazard ratios are per SD residual change in the cognitive measures. Cox regression models were performed only where there were at least 40 events. The power for detecting reliable differences is small with <40 events.

The covariates were chosen on the basis that each makes an independent contribution to mortality risk (39,40) and cognitive ability (41–48). Two models were fitted for each outcome: a) a demographic model including the corresponding HALS1 cognitive measure, age, and gender; b) a fully adjusted model controlling for the corresponding HALS1 cognitive measure, age, gender, social class, education, smoking status, alcohol consumption, physical activity, and three measures of health status (BMI, BP, and FEV). Analyses are carried out for all ages together and, where numbers permit, for broad age groups.

Overall performance on the RT and cognitive tasks did decline between HALS1 and HALS2. However, as a low value for the RT parameters and a high score on the memory and reasoning tests represent higher ability, an increase in the RT scores between HALS1 and HALS2 denotes decline in performance whereas a decrease in scores on the cognitive measures indicates decline in performance.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
CONCLUSIONS
 NOTES
 REFERENCES
 
Only those individuals with complete data for the RT parameters and cognitive measures and all covariates were included in the analysis. Of the 5352 individuals who were eligible, 3802 (1690 men, 2112 women) had complete information. Of these, 690 (376 men, 314 women) died during the 13 years of follow-up. The main causes of death were all CVDs, CHD, and all nonlung cancers (Table 1). Mean ± SD age of the sample was 43.95 ± 15.33 years (range = 18–87 years) at HALS1 and 50.92 ± 15.31 years (range = 25–94 years) at HALS2.


View this table:
[in this window]
[in a new window]

  		TABLE 1. Frequencies (%) for the Main Causes of Death

 

Mean scores for the RT parameters showed a slight, although nonsignificant, increase between HALS1 and HALS2— denoting a decline in performance (Table 2). There was no substantial change in the memory scores or reasoning scores between HALS1 and HALS2. Correlations among the regressed-change scores were low between the cognitive and reaction time measures (mean r = .1; range = .08–.18) and moderate-to-large between the RT measures themselves (mean r = .37; range = .23–0.57). As expected, the largest correlations were between scores derived from the same test, such as the regressed-change scores for SRT mean and simple RT variability (r = .57; p < .0001), and the regressed-change scores for CRT mean and CRT variability (r = .43; p < .0001). In contrast, correlations between the regressed-change scores and the covariates were small and many were nonsignificant. Only age, education, and lung function correlated at >0.1 with the regressed-change measures. We also examined whether the covariates were able to predict the various categories of mortality events in this sample. Only gender (mean hazard ratio (HR) across all categories of mortality = 2.02; range of HRs across all categories of mortality = 1.60–2.65), age (mean HR = 0.92; range of HRs = 0.91–0.93), smoking status (mean HR = 1.62; range of HRs = 1.25–2.78), and alcohol units consumed in a week (mean HR = 1.02; range of HRs = 1.01–1.02) significantly predicted death from each of the health outcomes.


View this table:
[in this window]
[in a new window]

  		TABLE 2. Mean ± Standard Deviation Scores for Each of the RT and Cognitive Measures at HALS1 and HALS2

 

All-Cause Mortality
For all ages together, significant associations between all-cause mortality and change in the RT and cognitive function measures were found in the demographic model (Table 3). Decline in cognitive performance was associated with a greater risk of mortality. The association was only slightly attenuated in the fully adjusted model.


View this table:
[in this window]
[in a new window]

  		TABLE 3. Risk of Death From All-Cause Mortality (690 Events) Associated With Cognitive Change: Hazard Ratios (95% Confidence Intervals) and p Values

 

Using this model, we examined whether joint prediction of the regressed-change cognitive measures showed additive effects. When entered jointly, there were significant effects for change in choice reaction variability, memory, and visual-spatial reasoning in the model. These effects were additive.

For the 40- to 59-year age group, a significant association between cognitive change and risk of death from all-causes in the demographic model was found for SRT mean, CRT mean, CRT variability, and reasoning (Table 3). Slight attenuation was noted in the fully adjusted model.

In the 60+ age group, change in all the RT parameters and cognitive measures were significantly associated with all-cause mortality in the demographic model (Table 3). The associations showed very slight attenuation in the fully adjusted model.

All Cardiovascular Disease
For the all ages group, a greater relative decline in SRT mean and variability, CRT mean and variability, and memory was associated with an increased risk of death from all CVDs in the demographic model (Table 4). The magnitude of the effect was mostly unaltered in the fully adjusted model.


View this table:
[in this window]
[in a new window]

  		TABLE 4. Risk of Death From All CVDs (307 Events) Associated With Cognitive Change: Hazard Ratios (95% Confidence Intervals) and p Values

 

For those aged between 40 and 59 years, risk of death from all CVDs was associated with change in all four of the RT parameters only (Table 4). In the fully adjusted model, there was very little attenuation for the SRT measures but some attenuation for the CRT measures.

In the 60+ age group, change on all cognitive tests, apart from reasoning, was significantly associated with death from all CVDs in the demographic model (Table 4). The magnitudes of the hazard ratios were mostly unaltered in the fully adjusted model.

Coronary Heart Disease
In the demographic model, when all ages were examined together, a significantly higher risk of death from CHD was found for change on all the RT parameters and memory (Table 5). In the fully adjusted model, the only test that showed some modest attenuation was CRT variability from a hazard ratio of 1.22 to 1.16. Nevertheless, the association remained significant.


View this table:
[in this window]
[in a new window]

  		TABLE 5. Risk of Death From CHD (165 Events) Associated With Cognitive Change: Hazard Ratios (95% Confidence Intervals) and p Values

 

In the 40- to 59-year age group, only change in the RT parameters was significantly associated with an increased risk of death from CHD (Table 5). Very little attenuation was noted in the fully adjusted model.

After controlling for age and gender, for those aged ≥60 years, risk of death from CHD was significantly associated with change in performance on SRT variability, CRT mean, and memory (Table 5). In the fully adjusted model, change in only these three tests remained significantly associated with death from CHD.

Stroke
Only change in the four RT parameters was significantly associated with an increased risk of death from stroke. There was very little attenuation after adding the remaining covariates to the model.

For those aged ≥60 years, only change in SRT mean and SRT variability and CRT mean were significantly associated with an increased risk of death from stroke (Table 6).


View this table:
[in this window]
[in a new window]

  		TABLE 6. Risk of Death From Stroke (78 Events) Associated With Cognitive Change: Hazard Ratios (95% Confidence Intervals) and p Values

 

Respiratory Disease
For the entire sample, change in all measures except SRT variability was significantly associated with an increased risk of death from respiratory disease in the demographic model (Table 7). However, in the fully adjusted model, the only measures that remained significantly associated with risk of death from respiratory disease were change in CRT mean, CRT variability, and memory.


View this table:
[in this window]
[in a new window]

  		TABLE 7. Risk of Death From Respiratory Disease (95 Events) Associated With Cognitive Change: Hazard Ratios (95% Confidence Intervals) and p Values

 

In the demographic model, change in CRT mean and variability, memory, and reasoning performance was significantly associated with risk of death from respiratory disease in those aged ≥60 years (Table 7). There was slight attenuation in the fully adjusted model.

Lung Cancer and All Nonlung Cancers
For the entire sample, there was no association between risk of death from lung cancer and change on any of the RT parameters or cognitive measures even in the demographic model (Table 8).


View this table:
[in this window]
[in a new window]

  		TABLE 8. Risk of Death From Lung Cancer (47 Events) Associated With Cognitive Change: Hazard Ratios (95% Confidence Intervals) and p Values

 

In the demographic model, the only significant association between cognitive change and risk of death from all nonlung cancers was for CRT mean and variability for the all ages group and the 60+ age group (Table 9). There was very little attenuation in the fully adjusted model.


View this table:
[in this window]
[in a new window]

  		TABLE 9. Risk of Death From All Nonlung Cancers (124 Events) Associated With Cognitive Change: Hazard Ratios (95% Confidence Intervals) and p Values

 


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 CONCLUSIONS
NOTES
 REFERENCES
 
Results suggest that increases in mean and variability of SRT and CRT and decreases in memory and reasoning performance are associated with an increased risk of all-cause mortality and death from all CVDs, CHD, stroke, and respiratory disease. Significant associations with all-cause mortality were noted for those aged 40 to 59 years and >60 years. For those aged between 40 and 59 years, it was only possible to examine associations with all CVDs and CHD. Only decline on the four RT parameters were associated with increased risk of death from these two diseases. For those aged ≥60 years, risk of death from all CVDs was associated with decline in all RT and memory in the fully adjusted model. As these associations remained after controlling for baseline values of the same cognitive measure plus a range of covariates, it is suggested that cognitive decline across a 7-year period is an independent predictor of mortality across the subsequent 13 years.

The findings that increases in both SRT and CRT mean and variability are associated with an increased risk of all-cause and cause-specific mortality are novel. No study to date has used RT change to examine the relationship between cognitive decline and mortality. Few of the past studies on cognitive decline and mortality have considered the role of processing speed as a potential explanation (3,25). However, it has been found that tests of processing speed, such as psychomotor speed (49), are more sensitive to age-associated declines in cognition than other more complex cognitive tasks (50,51). Hertzog et al. (52) found that decline in psychomotor speed was significantly associated with CVD. These associations between processing speed and CVD also support the significant associations between RT and risk of death from all CVDs and CHD noted in the present study.

The literature on decline in memory and reasoning ability and mortality is mixed. The findings of the current study are supported by Deeg et al. (8) who used the Wechsler Memory scale, Johansson and Berg (11) who used Digit Span, and Colsher and Wallace (53) who used a 20-word list learning test of immediate recall. All three studies found decline in memory to be a significant predictor of all-cause mortality. Berg (10) and Anstey et al. (5) found decline in reasoning to be predictive of later mortality. However, these studies were completed on participants aged >65 years. Therefore, finding an association between decline in reasoning and increased risk of mortality in adults aged 40 to 59 years is novel.

However, there are methodological issues. As expected, there were few deaths in the 20- to 39-year age group. Consequently, we were not able to examine any associations in this group. Nevertheless, the ability to include middle-aged adults in our analysis is unique within the literature. Having a younger age group, assessing the impact of change, and controlling for baseline cognition reduces the likelihood that the association is due to preexisting pathology. Furthermore, the extensive range of data in HALS enabled us to control for a number of known risk factors for CVD. However, it is important to consider selection bias. The majority of lost data were due to individuals not taking part in HALS2, whereas those who did return for HALS2 were healthier. Therefore, the exclusion of less healthy individuals may have diluted the true effects. Likewise, practice effects have been found to persist over periods up to 7 years (21,54,55). Any practice effects in the current study would also have underestimated decline and therefore its effect.

Mechanisms for the association between cognitive decline and mortality are unclear. Findings from our previous work provided some support for the theory that psychometric intelligence may be a marker for general bodily integrity (1,2,56). Slower and more variable RTs were significantly associated with increased mortality rates and RTs have been used as an indicator of speed of information processing within the central nervous system (27), which is inclined to be more sensitive to changes with age longitudinally (50,57). Therefore, it may be that decline in cognition reflects a decline in general bodily health. However, the fact that we found strong associations between cognitive decline and mortality in middle-aged adults who are more likely to be free of illness suggests this is not likely to be the case. More definitive conclusions on possible mechanisms require a greater understanding of possible modifiers of the RT-mortality association such as genetic factors. There is increasing evidence for the association between cerebrovascular disease and cognition due to the link between hypertension and increased cognitive decline (39,58–60). Linking these factors may be a genetic predisposition, as apoliopoprotein E polymorphism has been found to influence cognition (61) and also CVD (62).

Our previous studies (1,2) found a cross-sectional association between RT and mortality. Here, we have been able to extend these findings and examine the association between cognitive decline and mortality. As the association showed very little attenuation after adding all the covariates to the model, it suggests that declines in RT are independent predictors of all-cause and cause-specific mortality.


   NOTES
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 CONCLUSIONS
 NOTES
REFERENCES
 
Received for publication September 29, 2006; revision received May 17, 2007.

This study was funded by a grant from the Scottish Executive: Chief Scientist Office. Ian Deary is the recipient of a Royal Society-Wolfson Research Merit Award.

DOI:10.1097/PSY.0b013e31814c3e7c


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 CONCLUSIONS
 NOTES
 REFERENCES
 

  1. Shipley BA, Der G, Taylor MD, Deary IJ. Cognition and all-cause mortality across the entire adult age range: health and lifestyle survey. Psychosom Med 2006;68:17–24.[Abstract/Free Full Text]
  2. Shipley BA, Der G, Taylor MD, Deary IJ. Cognition and mortality from the major causes of death: the health and lifestyle survey. J Psychosom Res. Submitted.
  3. Bosworth HB, Siegler IC. Terminal change in cognitive function: an updated review of longitudinal studies. Exp Aging Res 2002;28:299–315.[CrossRef][Medline]
  4. Korten AE, Jorm AF, Jiao Z, Letenneur L, Jacomb PA, Henderson AS, Christensen H, Rodgers B. Health, cognitive, and psychosocial factors as predictors of mortality in an elderly community sample. J Epidemiol Community Health 1999;53:83–8.[Abstract]
  5. Anstey KJ, Luszcz MA, Giles LC, Andrews GR. Demographic, health, cognitive, and sensory variables as predictors of mortality in very old adults. Psychol Aging 2001;16:3–11.[CrossRef][Medline]
  6. Kleemeier R. Intellectual changes in the senium. Proceedings of the Social Statistics Section of the American Statistical Association. Washington, DC: Statistical Association; 1962.
  7. Riegel KF, Riegel RM. Development, drop, and death. Dev Psychol 1972;6:306–19.[CrossRef]
  8. Deeg DJH, Hofman A, van Zonneveld RJ. The association between change in cognitive function and longevity in Dutch elderly. Am J Epidemiol 1990;132:973–82.[Abstract/Free Full Text]
  9. Palmore E, Cleveland W. Aging, terminal decline, and terminal drop. J Gerontol 1976;31:76–81.[Abstract/Free Full Text]
  10. Berg S. Intelligence and terminal decline. In: Maddox GL, Busse EW, editors. Aging: the universal human experience. New York: Springer; 1987.
  11. Johansson B, Berg S. The robustness of the terminal decline phenomenon: longitudinal data from the digit-span memory test. J Gerontol 1989;44:184–6.[CrossRef]
  12. Colsher PL, Wallace RB. Longitudinal application of cognitive function measures in a defined population of community-dwelling elders. Ann Epidemiol 1991;1:215–30.[Medline]
  13. Van der Wal EA, Sandman CA. Evidence for terminal decline in the event-related potential of the brain. Electroencephalogr Clin Neurophysiol 1992;83:211–6.[CrossRef][Medline]
  14. Mortensen EL, Kleven M. A WAIS longitudinal study of cognitive development during the life span from ages 50 to 70. Dev Neuropsychol 1993;9:115–30.
  15. Bosworth HB, Schaie KW. Survival effects in cognitive function, cognitive style, and sociodemographic variables in the Seattle longitudinal study. Exp Aging Res 1999;25:121–39.[CrossRef][Medline]
  16. Bosworth HB, Schaie KW, Willis SL. Cognitive and sociodemographic risk factors for mortality in the Seattle longitudinal study. J. Gerontol B Psychol Sci Soc Sci 1999;54:273–82.
  17. Hassing LB, Johansson B, Berg S, Nilsson SE, Pedersen NL, Hofer SM, McClearn G. Terminal decline and markers of cerebro- and cardiovascular disease: findings from a longitudinal study of the oldest old. J Gerontol B Psychol Sci Soc Sci 2002;57:268–76.
  18. Hassing LB, Johansson B, Berg S, Nilsson SE, Pedersen NL, Hofer SM, McClearn G. Terminal decline and markers of cerebro- and cardiovascular disease: findings from a longitudinal study of the oldest old. J Gerontol B Psychol Sci Soc Sci 2002;57:268–76.
  19. Wilson RS, Beckett LA, Bienias JL, Evans DA, Bennett DA. Terminal decline in cognitive function. Neurology 2003;60:1782–7.[Abstract/Free Full Text]
  20. Silberstein R, ed. Special section on death and cognition. Eur Psychologist 2006;11:161–235.[CrossRef]
  21. Rabbitt P, Lunn M, Wong D. Understanding terminal decline in cognition and risk of death: Methodological and theoretical implications of practice and dropout effects. Eur Psychologist 2006;11:164–71.[CrossRef]
  22. Sliwinski MJ, Stawski RS, Hall CB, Katz M, Verghese J, Lipton R. Distinguishing preterminal and terminal cognitive decline. Eur Psychologist 2006;11:172–81.[CrossRef]
  23. Anstey KJ, Mack HA, von Sanden C. The relationship between cognition and mortality in patients with stroke, coronary heart disease, or cancer. Eur Psychologist 2006;11:182–95.[CrossRef]
  24. Thorvaldsson V, Hofer SM, Johansson B. Aging and late-life terminal decline in perceptual speed: a comparison of alternative modelling approaches. Eur Psychologist 2006;11:196–203.[CrossRef]
  25. Ghisletta P, McArdle JJ, Lindenberger U. Longitudinal cognition-survival relations in old and very old age: 13-year data from the Berlin aging study. Eur Psychologist 2006;11:204–23.[CrossRef]
  26. Backman L, MacDonald SW. Death and cognition: synthesis and outlook. Eur Psychologist 2006;11:224–35.[CrossRef]
  27. Deary IJ, Der G, Ford G. Reaction times and intelligence differences; a population-based cohort study. Intelligence 2001;29:389–99.[CrossRef]
  28. Der G, Deary IJ. Age and sex differences in reaction time in adulthood: results from the United Kingdom health and lifestyle survey. Psychol Aging 2006;21:62–73.[CrossRef][Medline]
  29. Deary IJ, Der G. Reaction time, age, and cognitive ability: longitudinal findings from age 16 to 63 in representative population samples. Aging Neuropsychol Cognition 2005;12:187–215.[CrossRef]
  30. Rabbitt P, Osman P, Moore B, Stollery B. There are stable individual differences in performance variability, both from moment to moment and from day to day. Q J Exp Psychol A 2001;54:981–1003.[CrossRef][Medline]
  31. Heim AW: AH4 group test of general intelligence. UK: NFER; 1970.
  32. Bosworth HB, Schaie KW. Survival effects in cognitive function, cognitive style, and sociodemographic variables in the Seattle Longitudinal study. Exp Aging Res 1999;25:121–39.[CrossRef][Medline]
  33. Hassing LB, Johansson B, Berg S, Nilsson SE, Pedersen NL, Hofer SM, McClearn G. Terminal decline and markers of cerebro- and cardiovascular disease: findings from a longitudinal study of the oldest old. J Gerontol B Psychol Sci Soc Sci 2002;57:268–76.
  34. Colsher PL, Wallace RB. Longitudinal application of cognitive function measures in a defined population of community-dwelling elders. Ann Epidemiol 1991;1:215–30.[Medline]
  35. Cox BD, Blaxter M, Buckle ALJ, Fenner NP, Golding JF, Gore M, Huppert FA, Nickson J, Roth M, Stark J, Wadsworth MEJ, Whichelow M. The health and lifestyle survey: a preliminary report. London: The Health Promotion Research Trust; 1987.
  36. Cox BD, Whichelow M. Sample structure, data collection and tracing procedures. In: Cox BD, Huppert FA, Whichelow M, editors. The health and lifestyle survey: seven years on. Aldershot: Dartmouth; 1993.
  37. Cox BD, Huppert F, Whichelow MJ. The health and lifestyle survey: seven years on. Aldershot, England: Dartmouth Publishing Company Limited; 1993.
  38. Campbell DT, Kenny DA. A primer on regression artifacts. New York: Guilford Press; 2003.
  39. Starr JM, Taylor MD, Hart CL, Davey SG, Whalley LJ, Hole DJ, Wilson V, Deary IJ. Childhood mental ability and blood pressure at midlife: linking the Scottish Mental Survey 1932 and the Midspan studies. J Hypertens 2004;22:893–7.[CrossRef][Medline]
  40. Taylor MD, Hart CL, Davey SG, Starr JM, Hole DJ, Whalley LJ, Wilson V, Deary IJ. Childhood mental ability and smoking cessation in adulthood: prospective observational study linking the Scottish mental survey 1932 and the Midspan studies. J Epidemiol Community Health 2003;57:464–5.[Free Full Text]
  41. Neisser U, Boodoo G, Bouchard T, Boykin A, Brody N, Ceci S, Halpern D, Loehlin J, Perloff R, Sternberg R, Urbiba S. Intelligence: knowns and unknowns. Am Psychol 1996;51:77–101.[CrossRef]
  42. Anstey KJ, Windsor TD, Jorm AF, Christensen H, Rodgers B. Association of pulmonary function with cognitive performance in early, middle and late adulthood. Gerontology 2004;50:230–4.[CrossRef][Medline]
  43. Halkjaer J, Holst C, Sorensen TI. Intelligence test score and educational level in relation to BMI changes and obesity. Obes Res 2003;11:1238–45.[Medline]
  44. Richards M, Hardy R, Wadsworth ME. Does active leisure protect cognition? Evidence from a national birth cohort. Soc Sci Med 2003;56:785–92.[CrossRef][Medline]
  45. Richards M, Jarvis MJ, Thompson N, Wadsworth ME. Cigarette smoking and cognitive decline in midlife: evidence from a prospective birth cohort study. Am J Public Health 2003;93:994–8.[Abstract/Free Full Text]
  46. Whalley LJ, Fox HC, Deary IJ, Starr JM. Childhood IQ, smoking, and cognitive change from age 11 to 64 years. Addict Behav 2005;30:77–88.[CrossRef][Medline]
  47. Kalmijn S, Van Boxtel MP, Verschuren MW, Jolles J, Launer LJ. Cigarette smoking and alcohol consumption in relation to cognitive performance in middle age. Am J Epidemiol 2002;156:936–44.[Abstract/Free Full Text]
  48. Glynn RJ, Beckett LA, Hebert LE, Morris MC, Scherr PA, Evans DA. Current and remote blood pressure and cognitive decline. JAMA 1999;281:438–45.[Abstract/Free Full Text]
  49. Salthouse TA. Speed mediation of adult age differences in cognition. Dev Psychol 1993;29:722–38.[CrossRef]
  50. Schaie KW. The hazards of cognitive aging. Gerontologist 1989;29:484–93.[Abstract]
  51. Elias MF, Robbins MA, Elias PK. A 15-year longitudinal study of Halstead-Reitan neuropsychological test performance. J Gerontol B Psychol Sci Soc Sci 1996;51:331–4.
  52. Hertzog C, Schaie KW, Gribbin K. Cardiovascular disease and changes in intellectual functioning from middle to old age. J Gerontol 1978;33:872–83.[Abstract]
  53. Colsher PL, Wallace RB. Longitudinal application of cognitive function measures in a defined population of community-dwelling elders. Ann Epidemiol 1991;1:215–30.[Medline]
  54. Rabbitt P, Diggle P, Smith D, Holland F, Mc IL. Identifying and separating the effects of practice and of cognitive ageing during a large longitudinal study of elderly community residents. Neuropsychologia 2001;39:532–43.[CrossRef][Medline]
  55. Rabbitt P, Diggle P, Holland F, McInnes L. Practice and drop-out effects during a 17-year longitudinal study of cognitive aging. J Gerontol B Psychol Sci Soc Sci 2004;59:84–97.
  56. Whalley LJ, Deary IJ. Longitudinal cohort study of childhood IQ and survival up to age 76. BMJ 2001;322:819.[Abstract/Free Full Text]
  57. Elias MF, Robbins MA, Elias PK. A 15-year longitudinal study of Halstead-Reitan neuropsychological test performance. J Gerontol B Psychol Sci Soc Sci 1996;51:331–4.
  58. Launer LJ, Masaki K, Petrovitch H, Foley D, Havlik RJ. The association between midlife blood pressure levels and late-life cognitive function. The Honolulu-Asia Aging Study. JAMA 1995;274:1846–51.[Abstract/Free Full Text]
  59. Farmer ME, Kittner SJ, Abbott RD, Wolz MM, Wolf PA, White LR. Longitudinally measured blood pressure, antihypertensive medication use, and cognitive performance: the Framingham Study. J Clin Epidemiol 1990;43:475–80.[CrossRef][Medline]
  60. Tzourio C, Dufouil C, Ducimetiere P, Alperovitch A. Cognitive decline in individuals with high blood pressure: a longitudinal study in the elderly. EVA study group. Epidemiology of vascular aging. Neurology 1999;53:1948–52.[Abstract/Free Full Text]
  61. Deary IJ, Whiteman MC, Pattie A, Starr JM, Hayward C, Wright AF, Carothers A, Whalley LJ. Cognitive change and the APOE epsilon 4 allele. Nature 2002;418:932.[CrossRef][Medline]
  62. Smith JD. Apolipoproteins and aging: emerging mechanisms. Ageing Res Rev 2002;1:345–65.[CrossRef][Medline]




This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrowRequest Permissions
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Shipley, B. A.
Right arrow Articles by Deary, I. J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Shipley, B. A.
Right arrow Articles by Deary, I. J.
Related Collections
Right arrow Other Epidemiology
Right arrow Cognitive Functioning

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS