International Journal of Epidemiology, 2014, 1874–1883
doi: 10.1093/ije/dyu170
Advance Access Publication Date: 22 August 2014
Original article
Miscellaneous
The influence of smoking, sedentary lifestyle
and obesity on cognitive impairment-free
life expectancy
Kaarin Jane Anstey,1* Andrew Kingston,2 Kim Matthew Kiely,1
Mary Alice Luszcz,3 Paul Mitchell4 and Carol Jagger2
1
Centre for Research on Ageing, Health and Wellbeing, The Australian National University, Canberra,
Australia, 2Institute for Health and Society, Newcastle University, Newcastle, UK, 3School of
Psychology, Flinders University, Adelaide, SA, Australia and 4Centre for Vision Research, Westmead
Millennium Institute and Department of Ophthalmology, University of Sydney, Sydney, NSW, Australia
*Corresponding author. Centre for Research on Ageing, Health and Wellbeing, The Australian National University,
Canberra, 0200, ACT Australia. E-mail: [email protected]
Accepted 24 July 2014
Abstract
Background: Smoking, sedentary lifestyle and obesity are risk factors for mortality and
dementia. However, their impact on cognitive impairment-free life expectancy (CIFLE)
has not previously been estimated.
Methods: Data were drawn from the DYNOPTA dataset which was derived by harmonizing and pooling common measures from five longitudinal ageing studies. Participants
for whom the Mini-Mental State Examination was available were included (N ¼ 8111,
48.6% men). Data on education, sex, body mass index, smoking and sedentary lifestyle
were collected and mortality data were obtained from Government Records via data linkage. Total life expectancy (LE), CIFLE and years spent with cognitive impairment (CILE)
were estimated for each risk factor and total burden of risk factors.
Results: CILE was approximately 2 years for men and 3 years for women, regardless of
age. For men and women respectively, reduced LE associated with smoking was 3.82
and 5.88 years, associated with obesity was 0.62 and 1.72 years and associated with
being sedentary was 2.50 and 2.89 years. Absence of each risk factor was associated
with longer LE and CIFLE, but also longer CILE for smoking in women and being sedentary in both sexes. Compared with participants with no risk factors, those with 2þ had
shorter CIFLE of up to 3.5 years depending on gender and education level.
Conclusions: Population level reductions in smoking, sedentary lifestyle and obesity
increase longevity and number of years lived without cognitive impairment. Years lived
with cognitive impairment may also increase.
Key words: Life expectancy, cognitive impairment, disability, healthy life, risk factors, DYNOPTA, smoking, sedentary, obesity
C The Author 2014; all rights reserved. Published by Oxford University Press on behalf of the International Epidemiological Association
V
1874
International Journal of Epidemiology, 2014, Vol. 43, No. 6
1875
Key Messages
• We used multistate models to analyse five pooled cohort studies to estimate life expectancies with and without cog-
nitive impairment.
• Australian men and women respectively spent approximately 2 and 3 years with cognitive impairment prior to death.
• Participants without risk factors of sedentary lifestyle, obesity and smoking had longer life expectancy and spent
more time without cognitive impairment. Increased longevity, however, was also associated with increased time with
cognitive impairment.
• Our results suggest that late-life cognitive impairment may be delayed through risk reduction, but previous estimates
of the effect of risk reduction on dementia prevalence may be overly optimistic, due to the competing risk of
longevity.
Introduction
As life expectancy (LE) continues to increase globally,
there is an urgent need to establish that increased years of
life are spent in good health. Health expectancies (HE) are
a broad set of population health indicators that combine
information on morbidity and mortality.1 They were developed to address the question of whether the extra years
of LE are healthy ones and to provide a metric for the compression of morbidity.2,3
As age is the strongest risk factor for dementia,4 increasing LE will result in a greater number of older adults with
dementia. In the context of HE research, there have been
more estimates based on cognitive impairment rather than
dementia, mainly due to data availability. Typically the
Mini-Mental State Examination (MMSE) is used to define
cognitive impairment5 and this instrument has been widely
validated as a screening instrument for dementia.6 Previous
research from Europe,7–9 Canada10 and the USA11,12 has
shown that on average, approximately 5–10% of life after
age 65 is spent with cognitive impairment
Given the lack of treatment or cure for dementia, there
has been focus on primary and secondary prevention
through risk reduction. Although not entirely uncontroversial,13 it has recently been estimated that a 10–25% reduction in seven key risk factors could prevent 1.1–3.0 million
Alzheimer’s disease (the largest cause of dementia) cases
internationally. A crucial issue therefore for public health
policies is the extent to which cognitive impairment-free
life expectancy (CIFLE) can be extended, and years with
cognitive impairment (CILE) can be shortened, through
reducing risk factors associated with late-life dementia.
Three key modifiable risk factors for dementia that have
recently been the focus of modelling of dementia risk reduction include smoking,14 sedentary lifestyle15,16 and
obesity.17 Current smoking has been associated with approximately 80% increased risk of dementia,14 and a sedentary life style or engaging in no physical activity is
associated with around 25% increased risk of dementia
compared with engaging in some physical activity.18
Overweight and obese body mass index (BMI) in midlife
(40 to 60 years of age) have been associated with a 26%
and 64% increased risk of dementia in later life, respectively,19 although the risk associated with late-life obesity
and overweight is less clear. Although these are known risk
factors for mortality, to our knowledge there has been no
previous investigation of their influence on cognitive impairment-free life expectancy.
To date the main focus of research has been on social
inequalities in CIFLE, specifically education as a proxy for
socioeconomic status,9 reflecting social inequalities in
broader measures of healthy LE.20 Low levels of education
have been associated with increased risk of dementia21 and
more years of life lived with cognitive impairment.9,11 Low
education is a marker of low socioeconomic status, and
this is also associated with a greater number of health risk
behaviours22 and so may act as a proxy for these, in addition to reflecting educational opportunities that may influence opportunities to develop cognitive reserve.23
The present study aimed to explore the extent to which
three modifiable risk factors for dementia, specifically
smoking, sedentary lifestyle and obesity, affected LE with
and without cognitive impairment in Australian men and
women, taking account of educational level.
Methods
Sample
The sample was drawn from DYNOPTA, a pooled dataset
of Australian Longitudinal Studies.24 For this study, contributing DYNOPTA studies had to include a common cognitive outcome measure and data on mortality. The
contributing studies included the Australian Longitudinal
Study of Ageing (ALSA),25 the Blue Mountains Eye Study
(BMES),26 the Canberra Longitudinal Study (CLS),27 the
Sydney Older Persons Study (SOPS)28 and the Path
1876
International Journal of Epidemiology, 2014, Vol. 43, No. 6
Through Life Study (PATH).29 ALSA is a random sample
drawn from the city of Adelaide in South Australia. BMES
is a random sample drawn from the Blue Mountain region
in New South Wales (NSW), encompassing two post code
areas, and is the only study to sample a semi-urban population that is not from a capital city. CLS comprises a random sample drawn from the city of Canberra. SOPS
comprises a random sample drawn from the inner city of
Sydney, NSW. PATH comprises a random sample drawn
from Canberra and Queanbeyan, NSW.The age ranges and
occasions of measurement for each study are reported in
Table 1. For all studies, participants were excluded from
analyses if they did not provide MMSE data at baseline
and most of these cases did not have MMSE data at later
waves. The total sample comprised 8111 participants
(3947 males).
(MRC-CFAS), analyses were also conducted with a cut
point of 21/22.9
Level of education was coded as a binary variable with
those leaving school before the age of 15 years classified as
‘early school leavers’ (ESL) and those leaving school aged
15 or older classified as ‘late school leavers’ (LSL). The
choice of education variables was determined by the need
to harmonize education across multiple studies. In
Australia during the time of the cohorts studied, a major
exit point from school occurred after 9 years of schooling,
at age 14.32 Baseline smoking status was coded as current,
former and never smoker. Self-reported frequency of walking sessions per week was used as a measure of physical activity, with sedentary behaviour defined by less than one
walking session per week. Obesity was defined by a BMI
of 30 or greater.
Measures
Statistical analyses
MMSE was used to assess probable cognitive impairment.30 A cut point of 23/24 (MMSE score 23 ¼ ‘probable dementia’; MMSE score >23 ¼ ‘no cognitive
impairment’) was used as a proxy for dementia status, as
reported previously.31 Clinical diagnoses of dementia were
available for two contributing studies and the 23/24 cut
point yielded the optimal sensitivity and specificity for
probable dementia which were 93.06% [confidence interval (CI): 90.66–94.89%] and 70.20% (CI: 69.09–
71.28%), respectively.31 To enable comparison of LE,
CIFLE and CILE by education with the Medical
Research Council Cognitive Function and Ageing Study
To assess the impact of health behaviours on cognitive impairment and death through CIFLE, we fitted a continuous
time multistate model (irreversible illness-death model)
with three states: no cognitive impairment, cognitive
impairment and death (absorbing state). This model estimated the instantaneous rate of transition (transition intensities) between the states, making the assumption that
transition from cognitive impairment to no cognitive impairment was not possible, and therefore any observed
transitions were due to recording error. We did not assume
that transition intensities were constant across observed
time intervals, but used a piecewise-constant intensity
Table 1. Characteristics of the contributing studies in DYNOPTA
Characteristics
Baseline year
MMSE wave 1
date
age range
n
MMSE wave 2
date
age range
n
MMSE wave 3
date
age range
n
MMSE wave 4
date
age range
n
ALSA
BMES
CLS
PATH
SOPS
1992
1992
1990
2001
1991
1992
65–103
2008
1997–2000
59–99
1962
1990
70–100
994
2001–2002
60–66
2547
1991–1993
75–98
600
1994–1995
67–106
1542
2001–2004
65–100
1410
1994
74–102
629
2005–2006
64–70
2185
1994–1996
78–99
426
2000–2001
73–102
652
1998
78–102
371
1997–1999
81–101
271
2003–2004
76–102
406
2002
82–106
209
2001–2003
84–106
56
ALSA, the Australian Longitudinal Study of Ageing; BMES:,the Blue Mountains Eye Study; CLS, the Canberra Longitudinal Study; PATH, the Personality and
Total Health Through Life Study; SOPS, the Sydney Older Persons Study
International Journal of Epidemiology, 2014, Vol. 43, No. 6
model with time from entry to the study as the timescale,
and age as an external time-dependent covariate, capturing
the time-dependency of the intensities.33 It is worth noting
that if covariates are time invariant (as in our study), this
model is equivalent to one with age as the timescale and
age as a covariate. We assessed the best-fitting model, in
terms of the length of time over which the intensities are
assumed constant, by a likelihood ratio test between a constant intensity model and the piecewise-constant intensity.
This resulted in the best-fitting time breakpoint being 6
months. The methodology analysing panel data using continuous-time multistate models with interval censoring has
been around for many years,34,35 albeit with the strong
assumption of time homogeneity. More recently other
methods have been developed in the R programming environment (www.r-project.org) based on: semi-parametric
and parametric approaches for the transition intensities;36
and the msm package37 with piecewise-constant intensities
which relaxes the time homogeneity assumption. To calculate life expectancies in different states we utilised the estimates from the piecewise-constant transition intensity
matrix in msm and used the method outlined by van den
Hout and Matthews33 with 1000 replications to produce
mean estimates of total LE, CIFLE and CILE (with corresponding standard errors) by sex, education and each health
behaviour. Models were first fitted for each health behaviour individually, and then combined into a single score
(0, 1, 2þ unhealthy behaviours); all models were adjusted
for education and stratified by sex. Where data were missing on covariates (health behaviours or education) they
were excluded. Participants with missing information on
cognitive state (other than the first observation) were retained and treated as censored.
Results
Description of sample
Table 2 shows the number of men and women in the sample at each follow-up according to cognitive impairment
status. At baseline 94.2% of men and 95.1% of women
had no cognitive impairment, 34.7% of men and 33.3% of
women left school at or before the age of 14, 10.7% of
men and 8.5% of women were smokers, 23.2% of men
and 27.0% of women were sedentary and 11.2% of men
and 15.7% of women were obese.
Overall life expectancies with and without
cognitive impairment for men and women
The estimated years lived with CILE increased slightly
with age (Figure 1) and formed a much greater proportion
1877
of remaining LE with increasing age (Supplementary
Table 1, available as Supplementary data at IJE online).
Men aged 65 spent 1.87 years (10.9%) of their remaining
17.04 years with cognitive impairment, and by age 85 the
estimated number of years with cognitive impairment for
men rose slightly to 2.13 years which amounted to 37.7%
of remaining life (5.6 years) (Supplementary Table 1, available as Supplementary data at IJE online). Compared with
men, women spent longer with cognitive impairment at all
ages, though given their longer LE, a similar proportion of
remaining life was spent with cognitive impairment: 2.95
years (13.9% of remaining 21.1 years) at age 65 and 3.09
years (39.5% of LE) at age 85. Comparison of estimates
using a cut point of 21/22 on the MMSE (see Supplementary Table 2 available as Supplementary data at IJE online)
with MRC-CFAS showed similar relatively constant years
with CILE regardless of age though years spent with cognitive impairment in MRC-CFAS were fewer for men
(around 1.4 years) and higher in women (around 2.5
years).
Years with and without cognitive impairment for
individual health risk factors
CIFLE and CILE are reported by each health risk factor individually in Table 3 (high education) and Supplementary
Table 3 (low education; available as Supplementary data
at IJE online). When the three health factors in higheducated individuals were viewed separately, smoking was
associated with the most years of life lost. At age 65 male
smokers lived 3.82 years (SE 0.74 years) less than male
non-smokers (P < 0.0001) whereas for females the difference was 5.88 years (SE 0.97 years, P < 0.0001).
Compared with non-smokers, CIFLE at age 65 for male
smokers was 2.95 years (SE 0.70 years) less and for female
smokers 4.60 years (SE 0.90 years) less (both P < 0.0001).
However, non-smokers were estimated to live for longer
with cognitive impairment in both absolute (years) and
relative terms (proportion of total LE). The difference in
CILE between non-smokers and smokers was just under
1 year for men (P ¼ 0.05) and just over 1 year for women
(P ¼ 0.02).
The difference in LE at age 65 between obese and nonobese participants was small (men: 0.62 years, SE 0.94
years, P ¼ 0.51; women: 1.72 years, SE 0.93 years,
P ¼ 0.06). Similarly obesity conferred only a small reduction in CIFLE at age 65 (men: 0.94 years, SE 0.79 years,
P ¼ 0.23; women: 1.11 years, SE 0.90 years, P ¼ 0.22) and
in years with cognitive impairment (Table 3). Finally,
those physically inactive men and women had shorter LE
at age 65 (men: 2.50 years, SE 0.74 years, P ¼ 0.0007;
women: 2.89 years, SE 0.89 years, P ¼ 0.001) and
–
–
–
–
–
–
–
–
–
–
–
1.2
0.9
0.7
–
–
–
–
–
–
–
–
–
–
3.7
7.4
11.1
(SD)
–
M
Years in
study
M ¼ mean; ALS ¼ age left school.
Baseline
Education
ALS 14
ALS >14
Missing
Smoking
Never/former
Current smoker
Missing
Obesity
Not obese
Obese
Missing
Physical activity
Active
Inactive
Missing
Follow-up
1st follow-up
2nd follow-up
3rd follow-up
Characteristics
2730
479
196
973
830
1915
2229
432
1057
3319
389
10
1225
2482
11
3718
N
69.2
12.1
5.0
24.7
21.0
48.5
56.5
10.9
26.8
84.1
9.9
0.3
31.0
62.9
0.3
94.2
(%)
No cognitive
impairment
159
75
67
85
86
58
101
13
115
195
30
4
147
79
3
229
N
4.0
1.9
1.7
2.2
2.2
1.5
2.6
0.3
2.9
4.9
0.8
0.1
3.7
2.0
0.1
5.8
(%)
Cognitive
impairment
566
1034
1348
–
–
–
–
–
–
–
–
–
–
–
–
–
N
14.3
26.2
34.2
–
–
–
–
–
–
–
–
–
–
–
–
–
(%)
Deceased
(cumulative)
Males (N ¼ 3947)
Table 2. Characteristics of the sample at baseline and follow-up, by sex
492
2359
2336
–
–
–
–
–
–
–
–
–
–
–
–
–
N
12.5
59.8
59.2
–
–
–
–
–
–
–
–
–
–
–
–
–
(%)
Missing
2969
629
294
1032
1040
1890
2195
640
1127
3,606
339
17
1265
2683
14
3962
N
71.3
15.1
7.1
24.8
25.0
45.4
52.7
15.4
27.1
86.6
8.1
0.4
30.4
64.4
0.3
95.1
(%)
No cognitive
impairment
159
80
70
65
82
55
57
12
133
170
18
14
119
68
15
202
N
3.8
1.9
1.7
1.6
2.0
1.3
1.4
0.3
3.2
4.1
0.4
0.3
2.9
1.6
0.4
4.9
(%)
Cognitive
impairment
412
745
948
–
–
–
–
–
–
–
–
–
–
–
–
–
N
9.9
17.9
22.8
–
–
–
–
–
–
–
–
–
–
–
–
–
(%)
Deceased
(cumulative)
Females (N ¼ 4164)
624
2710
2852
–
–
–
–
–
–
–
–
–
–
–
–
–
N
15.0
65.1
68.5
–
–
–
–
–
–
–
–
–
–
–
–
–
(%)
Missing
1878
International Journal of Epidemiology, 2014, Vol. 43, No. 6
International Journal of Epidemiology, 2014, Vol. 43, No. 6
1879
Figure 1. Total life expectancy and cognitively impaired life expectancy for males and females.
significantly shorter CIFLE for men but not women (men:
1.54 years, SE 0.70 years, P ¼ 0.03; women: 1.15 years, SE
0.76 years, P ¼ 0.13). On the other hand, due to the gain
in LE, expected years lived with cognitive impairment at
age 65 was significantly greater for the physically active
(men: 0.95 years, SE 0.42 years, P ¼ 0.02; women: 1.73
years, SE 0.58 years, P ¼ 0.002) (Table 3).
Although all life expectancies were lower in those with
lower education compared with the higher education
group, the differences between those with and without
risk factors were similar and thus conclusions were the
same with respect to their associations with CIFLE and
CILE.
Cognitive health expectancies for levels of risk
factor burden
To evaluate the association between risk factor burden and
HE outcomes, we created a combined score with 0, 1, or
2þ risk factors. LE, CIFLE and CILE at age 65 were all significantly greater for individuals with zero risk burden
compared with those with 2þ risk factors. For LE this
amounted to a gain of over 4 years for men and over
5 years for women, with the greatest gain for women with
lower education (5.94 years). For CIFLE there were gains
of between 3.0 and 3.5 years. Compared with those with
2þ risk factors, men with no risk factors had around
1 year more of CILE regardless of level of education,
whereas women with no risk factors and lower education
gained more years with cognitive impairment (2.46 years,
SE 0.74 years) than those with higher education (1.56
years, SE 0.49 years) (Table 4).
Discussion
The present study is the first to evaluate the impacts of
smoking, sedentary lifestyle and obesity on CIFLE and several important results emerged. First, risk factors vary in
the extent to which they are associated with both increased
LE and reduced CIFLE, and also vary by gender, thus presenting a complex picture of the overall impact of risk factors on CIFLE. Consistently with the wider literature,
smoking conferred the largest reduction in CIFLE for men
and women but did not show much variation between
higher and lower education groups. Overall the findings
here suggest that risk reduction will delay the onset of cognitive impairment or dementia, but not necessarily prevent
it. On the contrary, risk reduction will increase longevity
and some will also increase years of life spent with cognitive impairment.
1880
International Journal of Epidemiology, 2014, Vol. 43, No. 6
Table 3. Total life expectancies (TLE) and life expectancy free of cognitive impairment (CIFLE) and with cognitive impairment
(CILE) in years (standard errors in parentheses), and as a proportion of TLE by individual risk factor (smoking, obesity, physical
inactivity) for males and females with high education, at age 65 years
TLE
Males
Smoking
Obesity
Inactive
Females
Smoking
Obesity
Inactive
CIFLE
CILE
Risk factor
Years
(SE)
Years
(SE)
%a
Years
(SE)
%a
Yes
No
Gain (þ) or loss () with no risk factor
Yes
No
Gain (þ) or loss () with no risk factor
Yes
No
Gain (þ) or loss () with no risk factor
13.48
17.30
3.82
16.95
17.57
0.62
15.52
18.02
2.50
0.66
0.33
0.74
0.83
0.45
0.94
0.50
0.55
0.74
12.32
15.27
2.95
15.02
15.96
0.94
13.91
15.45
1.54
0.64
0.29
0.70
0.79
0.45
0.79
0.47
0.52
0.70
91.4
88.3
1.16
2.03
0.87
1.93
1.61
0.32
1.61
2.56
0.95
0.40
0.21
0.45
0.51
0.21
0.55
0.24
0.34
0.42
8.6
11.7
Yes
No
Gain (þ) or loss () with no risk factor
Yes
No
Gain (þ) or loss () with no risk factor
Yes
No
Gain (þ) or loss () with no risk factor
15.49
21.37
5.88
18.97
20.69
1.72
19.37
22.26
2.89
0.88
0.41
0.97
0.76
0.53
0.93
0.54
0.71
0.89
14.25
18.85
4.60
17.52
18.63
1.11
17.63
18.78
1.15
0.82
0.36
0.90
0.75
0.49
0.90
0.52
0.55
0.76
92.0
88.2
1.24
2.53
1.29
1.45
2.06
0.61
1.74
3.47
1.73
0.48
0.31
0.57
0.31
0.27
0.41
0.25
0.52
0.58
8.0
11.8
88.6
90.8
89.6
85.7
92.4
90.0
91.0
84.4
11.4
9.2
10.4
14.2
7.6
10.0
9.0
15.6
a
Proportion of TLE.
In the present study, whereas smoking and sedentary
lifestyle had the expected negative effects on LE and cognitive impairment, the findings for obesity were more
complex, particularly for men. This was not entirely unexpected as previous literature has failed to find increased
risk of dementia associated with obesity in later life.38
However it is possible that results for obesity are confounded by methodological limitations. BMI declines
in the 6 years prior to diagnosis with dementia39 and
hence in studies of older adults where such participants
are retained in the sample, the effect of high BMI will
not be identifiable. A limitation with this literature has
been the use of a continuous BMI measure and an assumption of a linear relationship between BMI and dementia
risk, whereas the research on mid-life obesity has demonstrated an inverse-U-shaped association between categories
of BMI and late-life dementia risk.17 On the other hand,
there is also the ‘obesity paradox’40 in which obesity has
been associated with better survival in some clinical
groups.
In the broader literature on dementia risk reduction, the
combined population attributable risk of risk factors for
dementia has been used to produce estimates of the number of cases of dementia that could be prevented by removing these key risk factors.41–43 Some authors using this
method have assumed that the risk factors are additive
when using combined risk scores to estimate the potential
to reduce dementia prevalence.42 However, it is likely that
the current modelling in this domain may be overly optimistic because it has been conducted without consideration
of competing risk44 caused by risk reduction, namely
increased longevity.45 Age is one of the strongest risk factors for dementia,4,31 and common cardiovascular risk
factors for dementia are also associated with increased
mortality risk, in both younger adulthood46 and late
adulthood.47
The current study has implications for the increasingly
important area of statistical modelling of the impact of dementia risk reduction on future prevalence. Our findings
relate to years of cognitive impairment at the population
level, and do not estimate number of cases of dementia.
Nevertheless, cases of dementia and years lived with cognitive impairment are two dimensions of the same phenomenon. Our findings suggest that previous modelling of the
potential to reduce the impact of dementia through risk reduction using population attributable risk of individual
risk factors may be exaggerated because they do not take
into account that increased longevity (resulting from risk
reduction) is associated with increased dementia risk.41,42
The results presented here are not projections, but are
International Journal of Epidemiology, 2014, Vol. 43, No. 6
1881
Table 4. Total life expectancies (TLE) and life expectancy free of cognitive impairment (CIFLE) and with cognitive impairment
(CILE). Results are shown in years (standard errors in parentheses) and as a proportion of TLE by total number of risk factors
(smoking, obesity, physical inactivity) for males and females at age 65, by level of education
Number of risky behaviours
Males low education
2þ
1
None
Gain (þ) or loss () for none vs 2þ
risky behaviours
Males high education
2þ
1
None
Gain (þ) or loss () for none vs 2þ
risky behaviours
Females low education
2þ
1
None
Gain (þ) or loss () for none vs 2þ
risky behaviours
Females high education
2þ
1
None
Gain (þ) or loss () for none vs 2þ
risky behaviours
TLE
Years
(SE)
13.12
15.57
17.25
4.13
0.72
0.44
0.47
0.86
14.03
16.74
18.30
4.27
0.79
0.42
0.40
0.89
16.96
18.94
22.90
5.94
0.97
0.52
0.62
1.15
17.88
19.70
22.89
5.01
0.89
0.49
0.49
1.02
CIFLE
P
Years
(SE)
%
0.66
0.41
0.39
0.77
90.32
88.95
86.67
<0.0001
11.85
13.85
14.95
3.10
0.72
0.39
0.35
0.80
91.66
90.92
88.36
<0.0001
12.86
15.22
16.17
3.31
0.84
0.46
0.46
0.96
87.15
85.06
79.78
<0.0001
14.78
16.11
18.27
3.49
0.83
0.47
0.43
0.93
93.06
91.47
87.81
<0.0001
16.64
18.02
20.10
3.46
drawn from actual data from individuals in cohort studies
and hence are less subject to bias. Our findings of a relatively constant number of years with cognitive impairment,
regardless of age, confirm others, though study differences
in the absolute number of years may reflect cohort differences in education.
The results of the present study need to be interpreted in
the context of its limitations. All data on risk factors were
obtained by self-report. Certain lifestyle factors, particularly physical inactivity, had a large number of missing values. However this was due to this information not being
included in specific studies, rather than individual participant-level non-response. We undertook sensitivity analyses, repeating analyses for smoking and obesity as single
risk factors based on all studies with relevant data, and
found very similar values for life and health expectancies.
Data were not drawn from nationally representative samples, but all contributing datasets were population-based
and representative of the region from which they were
drawn. There was an under-representation of adults living
in residential facilities. The study did not take into account
CILE
P
Years
(SE)
%
0.37
0.23
0.29
0.47
9.68
11.05
13.39
0.0001
1.27
1.72
2.31
1.04
0.35
0.21
0.26
0.44
8.34
9.14
11.64
<0.0001
1.17
1.53
2.13
0.96
0.54
0.31
0.51
0.74
12.85
14.94
20.26
0.0003
2.18
2.83
4.64
2.46
0.37
0.22
0.32
0.49
6.94
8.48
12.23
0.0002
1.24
1.67
2.80
1.56
P
0.03
0.03
0.0009
0.0014
known genetic risk factors for dementia such as the apolipoprotein E genotype because these data were not available.48 Our analysis has focused on LE without cognitive
impairment and this is not the same as HLE, since individuals may be free of cognitive impairment but have other
conditions.
This study also had several strengths including a large
sample size. The statistical models used allowed for covariates and hence risk factors could be examined after adjusting for education, which has not been done previously.
Modelling was conducted on longitudinal data that had
been linked to the national death registry. The MMSE has
been used in previous studies of CIFLE and hence allows
the results of this study to be compared with findings from
other countries.9
Our findings clearly support the view that risk reduction will delay incident dementia. However, the impact on
life spent with cognitive impairment is less clear. Longevity
itself increases the risk of cognitive impairment, so there
appears to be somewhat of a longevity paradox with dementia risk reduction.
1882
International Journal of Epidemiology, 2014, Vol. 43, No. 6
Supplementary Data
Supplementary data are available at IJE online.
Funding
9.
10.
This work was supported by a National Health and Medical
Research Council grant # 410215 and by the Australian Research
Council Centre of Excellence in Population Ageing Research
(CE110001029). K.J.A is funded by NHMRC Fellowship
#1002560. C.J. is funded by the AXA Research Fund.
11.
Conflict of interest: None declared.
12.
Acknowledgements
The data on which this research is based were drawn from several
Australian longitudinal studies including: the Australian
Longitudinal Study of Ageing (ALSA), the Blue Mountain Eye Study
(BMES), the Canberra Longitudinal Study of Ageing (CLS), the
Personality And Total Health Through Life Study (PATH) and the
Sydney Older Persons Study (SOPS). These studies were pooled and
harmonized for the Dynamic Analyses to Optimize Ageing
(DYNOPTA) project. All studies would like to thank the participants for volunteering their time to be involved in the respective
studies. Details of all studies contributing data to DYNOPTA,
including individual study leaders and funding sources, are available
on the DYNOPTA website (http://DYNOPTA.anu.edu.au). The
findings and views reported in this paper are those of the author(s)
and not those of the original studies or their respective funding
agencies.
13.
14.
15.
16.
17.
References
1. Field M, Gold MR. Summarizing Population Health: Directions
for the Development and Application of Population Metrics.
Washington, DC: National Academy Press, 1998.
2. Fries JF. Aging, natural death, and the compression of morbidity.
N Engl J Med 1980;303:130–35.
3. Joly P, Durand C, Helmer C, Commenges D. Estimating life
expectancy of demented and institutionalized subjects from
interval-censored observations of a Multi-state model. Stat
Modelling 2009;9:345–60.
4. Jorm AF, Dear KB, Burgess NM. Projections of future numbers
of dementia cases in Australia with and without prevention. Aust
NZ J Psychiatry 2005;39:959–63.
5. Jagger C, Matthews R, Lindesay J, Robinson T, Croft P, Brayne
C. The effect of dementia trends and treatments on longevity and
disability: a simulation model based on the MRC Cognitive
Function and Ageing Study (MRC CFAS). Age Ageing
2009;38:319–25; discussion 251.
6. Cherbuin N, Anstey KJ, Lipnicki DM. Screening for dementia: a
review of self- and informant-assessment instruments. Int
Psychogeriatr 2008;20:431–58.
7. Jagger C, Matthews R, Melzer D, Matthews F, Brayne C.
Educational differences in the dynamics of disability incidence,
recovery and mortality: Findings from the MRC Cognitive
Function and Ageing Study (MRC CFAS). Int J Epidemiol
2007;36:358–65.
8. Melzer D, McWilliams B, Brayne C, Johnson T, Bond J.
Socioeconomic status and the expectation of disability in old
18.
19.
20.
21.
22.
23.
24.
25.
26.
age: estimates for England. J Epidemiol Community Health
2000;54:286–92.
Matthews FE, Jagger C, Miller LL, Brayne C. Education differences in life expectancy with cognitive impairment. J Gerontol A
Biol Sci Med Sci 2009;64:125–31.
Dubois MF, Hebert R. Cognitive-impairment-free life expectancy for Canadian seniors. Dement Geriatr Cogn Disord
2006;22:327–33.
Lievre A, Alley D, Crimmins EM. Educational differentials in life
expectancy with cognitive impairment among the elderly in the
United States. J Aging Health 2008;20:456–77.
Suthers K, Kim JK, Crimmins E. Life expectancy with cognitive
impairment in the older population of the United States.
J Gerontol B Psychol Sci Soc Sci 2003;58:S179–86.
Williams JW, Plassman BL, Burke J, Holsinger T, Benjamin S.
Preventing Alzheimer’s Disease and Cognitive Decline. Evidence
Report/Technology Assessment, No. 193. Rockville, MD:
Agency for Healthcare Research and Quality, 2010.
Anstey KJ, von Sanden C, Salim A, O’Kearney R. Smoking as a
risk factor for dementia and cognitive decline: a meta-analysis of
prospective studies. Am J Epidemiol 2007;166:367–78.
Bowen ME. A prospective examination of the relationship
between physical activity and dementia risk in later life. Am J
Health Promot 2012;26:333–40.
Morgan GS, Gallacher J, Bayer A, Fish M, Ebrahim S, BenShlomo Y. Physical activity in middle-age and dementia in later
life: findings from a prospective cohort of men in Caerphilly,
South Wales and a meta-analysis. J Alzheimers Dis 2012;31:
569–80.
Anstey KJ, Cherbuin N, Budge M, Young J. Body mass index in
midlife and late-life as a risk factor for dementia: a meta-analysis
of prospective studies. Obes Rev 2011;12:e426–37.
Scarmeas N, Luchsinger JA, Schupf N et al. Physical activity,
diet, and risk of Alzheimer disease. JAMA 2009;302:627–37.
Luck T, Luppa M, Briel S, Riedel-Heller SG. Incidence of mild
cognitive impairment: a systematic review. Dement Geriatr
Cogn Disord 2010;29:164–75.
Jagger C, Gillies C, Moscone F et al. Inequalities in healthy life
years in the 25 countries of the European Union in 2005: a crossnational meta-regression analysis. Lancet 2008;372:2124–31.
Caamano-Isorna F, Corral M, Montes-Martinez A, Takkouche
B. Education and dementia: a meta-analytic study.
Neuroepidemiology 2006;26:226–32.
Stringhini S, Dugravot A, Shipley M et al. Health behaviours,
socioeconomic status, and mortality: further analyses of the
British Whitehall II and the French GAZEL Prospective Cohorts.
PLoS Med 2011;8:e1000419.
Stern Y. Cognitive reserve. Neuropsychologia 2009;47:
2015–28.
Anstey KJ, Byles JE, Luszcz MA et al. Cohort profile: The
Dynamic Analyses to Optimize Ageing (DYNOPTA) project. Int
J Epidemiol 2009;39:44–51.
Luszcz MA, Bryan J, Kent P. Predicting episodic memory performance of very old men and women: contributions from age,
depression, activity, cognitive ability, and speed. Psychol Aging
1997;12:340–51.
Mitchell PSW, Attebo K, Healey PR. Prevalence of open-angle
glaucoma in Australia: the Blue Mountains Eye Study.
Ophthalmology 1996;103:1661–69.
International Journal of Epidemiology, 2014, Vol. 43, No. 6
27. Christensen H, Jorm AF, Henderson AS, Mackinnon AJ, Korten
AE, Scott LR. The relationship between health and cognitive
functioning in a sample of elderly people in the community. Age
Ageing 1994;23:204–12.
28. Broe GA, Creasey H, Jorm AF et al. Health habits and risk of
cognitive impairment and dementia in old age: a prospective
study on the effects of exercise, smoking and alcohol consumption. Aust N Z J Public Health 1998;22:621–23.
29. Anstey KJ, Christensen H, Butterworth P et al. Cohort profile:
the PATH through life project. Int J Epidemiol 2012;41:951–60.
30. Folstein MF, Robins LN, Helzer JE. The Mini-Mental State
Examination. Arch Gen Psychiatry 1983;40:812.
31. Anstey KJ, Burns RA, Birrell CL, Steel D, Kiely KM, Luszcz MA.
Estimates of probable dementia prevalence from populationbased surveys compared with dementia prevalence estimates
based on meta-analyses. BMC Neurol 2010;10:62.
32. Barcan A. A History of Australian Education. Melbourne, VIC:
Oxford University Press, 1980.
33. van den Hout A, Matthews FE. A piecewise-constant Markov
model and the effects of study design on the estimation of life
expectancies in health and ill health. Stat Methods Med Res
2009;18:145–62.
34. Kalbfleisch JD, Lawless JF. The analysis of panel data under a
Markov assumption. J Am Stat Assoc 1985;80:863–71.
35. Satten GA, Longini IM Jr. Markov chains with measurement
error: estimating the ‘true’ course of a marker of the progression
of human immunodeficiency virus disease. J R Stat Soc Series C
(Appl Stat) 1996;45:275–309.
36. Touraine C, Helmer C, Joly P. Predictions in an illness-death
model. Stat Methods Med Res 22 May 2013. Epub ahead of
print. PMID: 23698869.
37. Jackson CH. Multi-state models for panel data: The msm package for R. J Stat Softw 2011;38:1–29.
1883
38. Dahl AK, Lopponen M, Isoaho R, Berg S, Kivela SL. Overweight
and obesity in old age are not associated with greater dementia
risk. J Am Geriatr Soc 2008;56:2261–66.
39. Johnson DK, Wilkins CH, Morris JC. Accelerated weight loss
may precede diagnosis in Alzheimer disease. Arch Neurol
2006;63:1312–17.
40. Greenberg JA. The obesity paradox in the US population. Am J
Clin Nutr 2013;97:1195–200.
41. Ritchie K, Carriere I, Ritchie CW, Berr C, Artero S, Ancelin ML.
Designing prevention programmes to reduce incidence of dementia: prospective cohort study of modifiable risk factors. BMJ
2010;341:c3885.
42. Barnes DE, Yaffe K. The projected effect of risk factor reduction
on
Alzheimer’s
disease
prevalence. Lancet
Neurol
2011;10:819–28.
43. Dodge HH, Chang CC, Kamboh IM, Ganguli M. Risk of
Alzheimer’s disease incidence attributable to vascular disease in
the population. Alzheimer Dement 2011;7:356–60.
44. Andersen P, Geskus R, de Witte T, Putter H. Competing risks in
epidemiology: possibilities and pitfalls. Int J Epidemiol
2012;41:861–70.
45. Kryscio RJ, Abner EL, Lin Y et al. Adjusting for mortality when
identifying risk factors for transitions to mild cognitive impairment and dementia. J Alzheimer Dis 2013;35:823–32.
46. Saydah S, Bullard KM, Imperatore G, Geiss L, Gregg EW.
Cardiometabolic risk factors among US adolescents and young
adults and risk of early mortality. Pediatrics 2013;131:e679–86.
47. Huffman MD, Lloyd-Jones DM, Ning H et al. Quantifying options for reducing coronary heart disease mortality by 2020.
Circulation 2013;127:2477–84.
48. McKay GJ, Silvestri G, Chakravarthy U et al. Variations in apolipoprotein E frequency with age in a pooled analysis of a large
group of older people. Am J Epidemiol 2011;173:1357–64.