Accepted: 11 October 2016

DOI: 10.1111/ane.12709

ORIGINAL ARTICLE

Cognitive function in stroke survivors: A 10-­year

follow-­up study

H. Delavaran1,2 | A.-C. Jönsson2,3 | H. Lövkvist1,4 | S. Iwarsson3 | S. Elmståhl3,5 | 

B. Norrving1,2 | A. Lindgren1,2

1
Department of Clinical Sciences Lund,
Neurology, Lund University, Lund, Sweden
2
Department of Neurology and
Rehabilitation Medicine, Skåne University
Hospital, Lund, Sweden
3 of long-­term PSCI; (ii) compare two common cognitive assessment instruments; and (iii)
Department of Health Sciences, Lund
University, Lund, Sweden
4
Kliniska Studier Sverige - Forum
Söder, Skåne University Hospital, Lund,
Sweden
5
Department of Geriatrics, Skåne University
Hospital, Malmö, Sweden
Correspondence
H. Delavaran, Department of Neurology,
Skåne University Hospital, Lund University,
Lund, Sweden.
Email: hossein.delavaran@med.lu.se
Funding information
This study was supported by Region Skåne,
Lund University, the Swedish Heart and
Lung Foundation, the Freemasons Lodge of
Instruction EOS in Lund, King Gustaf V’s and
Queen Victoria’s Foundation, the Swedish
Stroke Association, and the Ribbingska
Foundation in Lund, Sweden.
Objectives: Post-­stroke cognitive impairment (PSCI) has considerable impact on pa-
tients and society. However, long-­term studies on PSCI are scarce and may be influ-
enced by assessment methods and selection bias. We aimed to (i) assess the prevalence
compare cognitive function of long-­term stroke survivors with non-­stroke persons.
Methods: Mini-­Mental State Examination (MMSE) and Montreal Cognitive Assessment
(MoCA) were administered to 10-­year survivors from a population-­based cohort of
first-­ever stroke patients included in the Lund Stroke Register, Sweden, in 2001-­2002.
PSCI was defined as MMSE<27 and/or MoCA<25 and severe cognitive impairment as
MMSE<23. Age-­ and sex-­matched non-­stroke control subjects who had performed
MMSE (but not MoCA) were recruited from the longitudinal population study “Good
Ageing in Skåne.” The odds of having cognitive impairment for stroke survivors com-
pared to controls were examined with logistic regression analyses adjusting for
education.
Results: Of 145 stroke survivors after 10 years, 127 participated. MMSE showed PSCI
in 46%, whereas MoCA displayed PSCI in 61%. Among the stroke survivors with
MoCA<25, 35% had MMSE≥27 (P<.001). The odds of having severe cognitive impair-
ment defined as MMSE<23 were higher among the stroke survivors compared to 354
controls (education-­adjusted; OR=2.5; P=.004).
Conclusions: Post-­stroke cognitive impairment was prevalent among 10-­year stroke
survivors, and the odds of having severe cognitive impairment were higher among the
stroke survivors compared to non-­stroke persons. The burden of long-­term PSCI
might have been underestimated previously, and MoCA may be more suitable than
MMSE to detect long-­term PSCI.

KEYWORDS
cerebrovascular diseases, dementia, mild cognitive impairment, strokes

1 |  INTRODUCTION
Post-­stroke cognitive impairment (PSCI), ranging from mild cognitive
impairment (MCI) to severe post-­stroke dementia (PSD), is associated
with increased mortality, disability, dependency, institutionalization
and higher costs of care.1-3 Thus, PSCI has a considerable impact on
patients, their families and healthcare resources. PSCI is likely to be-
come a mounting public healthcare challenge as the global burden of
stroke is huge and still growing.4
The overall prevalence of dementia in people aged ≥60 years
ranges between 6% and 8% in most regional estimates around the
world.5 However, there is substantial variation in the previously

Acta Neurol Scand 2016; 1–8 wileyonlinelibrary.com/journal/ane © 2016 John Wiley & Sons A/S.  |  1
Published by John Wiley & Sons Ltd
 |
2       DELAVARAN et al.
reported prevalence rates of PSD, ranging between 7% and 41%.6
When considering the differences in study designs, assessment meth-
ods and case mix, a systematic review showed that 10% of stroke
patients had dementia prestroke, an additional 10% developed de-
mentia shortly after stroke, and more than 30% had dementia after
6 3,7
recurrent stroke. In addition, MCI is at least as frequent as PSD.
However, most previous studies on cognitive function after stroke
1,3,6
are based on short-­term follow-­ups. The prevalence and charac-
teristics of PSCI in population-­based groups in a longer perspective,
for example at least 10 years after stroke onset is less studied, and
comparisons with non-­stroke individuals are particularly scarce. Two
previous community-­based studies reported a twofold increased risk
of dementia in long-­term follow-­up after stroke compared to non-­
stroke persons, but one of these studies was retrospective with base-
line assessments performed during 1960-­1984,8 whereas the other
9
only addressed PSD and not the whole spectrum of PSCI. A more
recent population-­based study reported a PSCI prevalence of 18%
after 10 years post-­event, but no comparison was made to non-­stroke
persons.10 Up-­to-­date and reliable data on the prevalence and charac-
teristics of long-­term PSCI is needed for a better understanding of the
overall burden of stroke and optimized healthcare planning.
Another concern is that various cognitive assessment methods
and diagnostic criteria exist, resulting in a lack of consensus on how
3,6,11
to define cognitive impairment. Brief tests of global cognitive
function are often used for clinical screening purposes and in research
12
studies. The Mini-­Mental State Examination (MMSE) is the most
commonly used cognitive screening test worldwide, but was origi-
nally developed to detect dementia and has several shortcomings, for
example weaknesses in detecting mild cognitive deficits and specific
cognitive problems frequently associated with stroke such as exec-
13-15
utive dysfunction. The Montreal Cognitive Assessment (MoCA)
was more recently developed to also detect MCI.16 Previous studies
have demonstrated a higher sensitivity of MoCA over MMSE for MCI
and executive function deficits, which suggests that MoCA could be
more suitable for cognitive assessment in stroke patients.17-20 As
long-­term PSCI (≥10 years after stroke) is scarcely studied, it is un-
clear whether the reported differences in the relative performance
of MMSE and MoCA also hold true for such long-­term post-­stroke
assessment.
We therefore examined a population-­based group of 10-­year
stroke survivors, aiming to (i) assess the prevalence and characteris-
tics of long-­term PSCI, measured with MMSE and MoCA; (ii) compare
the results from MMSE vs MoCA in long-­term stroke survivors; and
(iii) compare cognitive function, as measured with MMSE, between
long-­term stroke survivors and an age-­ and sex-­matched sample of
non-­stroke persons after adjustment for education.
2 |  METHODS
2.1 | Sample
Consecutive first-­ever stroke patients (n=416) were included pro-
spectively in the Lund Stroke Register (LSR) between March 1,
2001 and February 28, 2002. LSR is a population-­based stroke
study covering the catchment area of Skåne University Hospital
in Lund, Sweden, comprising eight municipalities with a total of
234 505 inhabitants (December 31, 2001). Multiple overlapping
sources were used to detect stroke patients, which have been pre-
viously described in depth.21 Stroke was defined by the established
World Health Organization criteria.22 Stroke severity at baseline
was assessed using the National Institutes of Health Stroke Scale
(NIHSS).23
2.2 | Ten-­year follow-­up and cognitive assessment
Follow-­up of all LSR stroke survivors (n=145) was organized 10 years
after stroke onset. The follow-­up procedure, as well as sample flow
from baseline to follow-­up, has been previously described in detail.24
The modified Rankin Scale (mRS; score 0-­5) was used to assess the
level of disability after 10 years.25 Vascular risk factors registered at
the 10-­year follow-­up comprised hypertension, atrial fibrillation, dia-
betes mellitus, hypercholesterolemia, and smoking (current and previ-
ous). Risk factors were defined on the basis of previous diagnosis and/
or ongoing relevant medication. The occurrence of recurrent stroke
was also registered.
A cognitive assessment session using MMSE15 and MoCA16 was
performed for each participating 10-­year stroke survivor. Stroke sur-
vivors were excluded if they were unwilling to participate; if their gen-
eral health condition precluded testing; or if they during the testing
situation were unable to complete the cognitive testing due to severe
dysphasia/aphasia, poor vision, or hearing impairment. The order of
testing was predetermined similar to previous studies, and each ses-
sion started with MMSE followed by MoCA.17 Identical tasks on the
MMSE and MoCA (serial 7s subtraction and orientation questions)
were only executed once. Test details are presented in Table S1.
The interpretation of the MMSE and MoCA scores was based
on recently recommended cutoffs derived from validations in stable
cerebrovascular cohorts.11,18,26 Hence, the MMSE scores were in-
terpreted according to the following recommended cutoffs, defining
cognitive impairment as follows: severe=0-­22 (indicative of demen-
tia), mild=23-­26 (indicative of MCI), or none=27-­30 (indicating nor-
mal cognitive function).11,18,26 The MoCA scores were interpreted
according to the following recommended cutoffs, categorizing cog-
nitive impairment as follows: severe=0-­19 (indicative of dementia),
mild=20-­24 (indicative of MCI), or none=25-­30 (indicating normal
cognitive function).11,18,26
We especially studied specific items of MMSE and MoCA that
involve executive functions, that is, the visuoconstruction subtest
of MMSE and the visuospatial/executive functions subtest of MoCA
which both assess visuoexecutive functioning20; as well as the verbal
fluency and abstraction tasks of MoCA which contain multiple aspects
of executive functions.16
Education level was registered as number of years of formal
schooling and categorized in accordance with the Swedish educa-
tion system: ≤9 years (grade school), 10-­12 years (high school), and
>12 years (college and/or university).
DELAVARAN et al.
2.3 | Control group
Non-­stroke control subjects were recruited from “Good Ageing in
Skåne” (GÅS), an ongoing longitudinal population study of aging and
care that started in 2001.27 The initial GÅS sample consisted of 2931
individuals. The participants in GÅS are aged ≥60 years, residents of
southern Sweden and have been included through random selection
from the National Population Register.27
For every stroke survivor aged ≥60 years in this study, three
age-­ and sex-­matched non-­stroke controls were selected from the
GÅS database by one of the co-­authors (blinded to other details re-
garding the stroke survivors). All controls had performed the same
MMSE version as the stroke survivors. Education level and vascular
risk factors, as defined above, were also registered for the controls.
2.4 | Ethics
This study was approved by the Regional Ethical Review Board in
Lund, Sweden. Informed consent was obtained from the study partici-
pants or their relatives.
2.5 | Statistical analysis
We graphically visualized and compared the MMSE vs MoCA scores of
the stroke survivors by use of a bubble plot.17 McNemar’s test was used
to analyze possible discrepancies between MMSE and MoCA at corre-
sponding cutoffs (MMSE<27 vs MoCA<25; MMSE<23 vs MoCA<20).
Mann-­Whitney’s two-­sample test and Fisher’s exact test were
used for univariate case-­control analyses. We also prespecified and
used simple logistic regression and multiple logistic regression analy-
ses, adjusting for the possible confounding effect of number of years
of education,14 to assess the odds of having any cognitive impair-
ment (MMSE<27) as well as severe cognitive impairment (MMSE<23)
among the stroke survivors compared to the controls.
P-­values <.05 were considered significant. All statistical calcula-
tions were performed using SPSS software (version 22, released 2013;
IBM SPSS Statistics for Windows, Armonk, NY: IBM Corp).
3 | RESULTS
Of the 145 stroke survivors after 10 years, 127 (87.6%) were included
in this study (Figure 1). All included stroke survivors completed the
testing with MMSE, and 122 (96.1%) also completed the MoCA (five
stroke survivors declined the subsequent MoCA testing). A total of
354 controls were included.
3.1 | Demographics and clinical characteristics
The stroke survivors had a higher educational attainment (median
8.5 years, range 5-­20) than the controls (median 7 years, range 4-­20;
Mann-­Whitney’s two-­sample test, P=.015). Details on demographics
are presented in Table 1.
|
      3
Among the 127 included stroke survivors, 111 (87.4%) had isch-
emic stroke at baseline, eight (6.3%) intracerebral hemorrhage, and
eight (6.3%) subarachnoid hemorrhage. Median NIHSS at baseline
was 3 (range 0-­27). In total, 17 (13.4%) stroke survivors had recurrent
stroke within 10 years.
At the 10-­year follow-­up, 96 (75.6%) stroke survivors had no or
only slight disability (mRS=0-­2), 17 (13.4%) had moderate disabil-
ity (mRS=3), and 14 (11.0%) had severe disability (mRS=4-­5). The
distribution of vascular risk factors after 10 years is also shown in
Table 1.
3.2 | MMSE and MoCA performance of the 10-­year
stroke survivors
The median MMSE score of the 127 stroke survivors was 27 (range
10-­30). Using MMSE, any cognitive impairment (MMSE<27) was
observed in 58 (45.7%) stroke survivors. Notably, 51 (40.2%) of the
stroke survivors failed the MMSE visuoconstruction subtest.
The median MoCA score of the 122 stroke survivors who com-
pleted the test was 23 (range 4-­30). According to MoCA, 75 (61.5%)
stroke survivors displayed any cognitive impairment (MoCA<25). With
regard to the MoCA items involving executive functions, no less than
94 (77.0%) stroke survivors had errors in the visuospatial/executive
functions subtest; 74 (60.7%) failed the verbal fluency task; and 69
(56.6%) had errors in the abstraction task.
The overall MMSE and MoCA scores of the study participants are
summarized in Table 2. Further details on the MMSE and MoCA sub-
test scores are presented in Tables S2 and S3, respectively.
3.3 | Comparison of MMSE with MoCA
Among the 122 stroke survivors who completed both tests, MMSE
scores were skewed toward higher values (median 27, interquartile
range 25-­29) compared to MoCA (median 23, interquartile range
19-­26).
In total, 49 (40.2%) stroke survivors were classified as cognitively
impaired by both tests (MMSE<27 and MoCA<25). Of 75 (61.5%)
stroke survivors with MoCA<25, 26 had normal MMSE. Conversely,
only four stroke survivors with normal MoCA scored MMSE<27.
These discrepancies reached statistical significance (McNemar’s test,
P<.001).
With regard to severe cognitive impairment (MoCA<20;
MMSE<23), 35 (28.7%) stroke survivors scored MoCA<20 of whom
16 had MMSE≥23. Only two stroke survivors with MMSE<23 scored
MoCA≥20. These discrepancies also reached statistical significance
(McNemar’s test, P=.001). Figure 2 illustrates the comparison of
MMSE vs MoCA.
3.4 | MMSE performance of the 10-­year stroke
survivors vs controls
The odds of having severe cognitive impairment defined as MMSE<23
were higher among stroke survivors as compared to controls (crude
 |
4       DELAVARAN et al.
OR=1.75; 95% CI: 1.00-­3.05; P=.048). The odds of having severe
cognitive impairment were even higher among the stroke survivors
when adjusting for education with multiple logistic regression analysis
(OR=2.48; 95% CI: 1.34-­4.59; P=.004). No such difference was seen
between the stroke survivors and the controls with regard to the odds
of having any cognitive impairment defined as MMSE<27 (education-­
adjusted; OR=1.03; 95% CI: 0.66-­1.60; P=.91). Table 3 displays the
simple and multiple logistic regression analyses, assessing the odds of
having impaired MMSE performance among the stroke survivors as
compared to the controls.
F I G U R E   1   Patient flowchart
Regarding visuoexecutive functioning, univariate analysis showed
that 50 (42.4%) stroke survivors failed the visuoconstruction subtest
compared to 57 (16.1%) controls (Fisher’s exact test, P<.001).
4 | DISCUSSION
Our study provides new population-­based data demonstrating a high
prevalence of PSCI (46% according to MMSE and 61% according
to MoCA) among 10-­year stroke survivors, and that visuoexecutive
DELAVARAN et al. |
      5

T A B L E   1   Demographics and vascular

Stroke survivors Control subjects
risk factors at 10-­year follow-­up
a b
(n=127) (n=118) (n=354)
Sex, n (%)
Men 73 (57.5) 66 (55.9) 198 (55.9)
Women 54 (42.5) 52 (44.1) 156 (44.1)
Age (years)
Median (range) 77 (27-­93) 78 (60-­93) 78 (60-­93)
Categories, n (%)
<60 9 (7.1) 0 0
60-­69.9 29 (22.8) 29 (24.6) 80 (22.6)
70-­79.9 39 (30.7) 39 (33.1) 116 (32.8)
80-­89.9 43 (33.9) 43 (36.4) 137 (38.7)
≥90 7 (5.5) 7 (5.9) 21 (5.9)
Education (years)
Median (range) 9 (5-­20) 8.5 (5-­20) 7 (4-­20)
Categories, n (%)
≤9 68 (53.5) 67 (56.8) 236 (66.7)
10-­12 20 (15.7) 17 (14.4) 52 (14.7)
>12 33 (26.0) 29 (24.6) 45 (12.7)
Vascular risk factors, n (%)
Hypertension 97 (76.4) 94 (79.7) 34 (9.6)
Atrial fibrillation 30 (23.6) 30 (25.4) 45 (12.7)
Diabetes mellitus 25 (19.7) 23 (19.5) 27 (7.6)
Hypercholesterolemia 76 (59.8) 72 (61.0) 47 (13.3)
Smoking (current and previous) 71 (55.9) 67 (56.8) 169 (47.7)
a
All 10-­year stroke survivors included.
b
All 10-­year stroke survivors aged ≥60 years and for whom controls were available.

deficits can be observed among no less than 77% of this group of
stroke survivors. The odds of having severe cognitive impairment de-
fined as MMSE<23 were 2.5 times higher among 10-­year stroke sur-
vivors as compared to the controls.
The prevalence of PSCI long-­term after stroke has been scarcely
studied, and the definitions of which levels of cognitive impairment
to include vary, and are not always explicit. In a recent population-­
based study from the South London Stroke Register, the prevalence
of PSCI (defined as MMSE<24 or Abbreviated Mental Test<8) was es-
timated at 18% after 10 years post-­event.10 When applying the same
MMSE cutoff, our results were similar (MMSE<24; 23%). Even though
MMSE<24 has high specificity and good sensitivity for detecting mod-
erate/severe cognitive impairment,12 it has low sensitivity for MCI.18
In fact, MMSE has relatively low sensitivity for MCI even at higher
18
cutoffs such as MMSE<27. Consequently, many cases with MCI may
not have been included in the South London Stroke Register estimate.
In contrast, MoCA<25 detects MCI with high sensitivity and accept-
able specificity.18 Hence, by including the entire range from MCI to
severe PSD, our findings suggest that PSCI among 10-­year stroke sur-
vivors may be more prevalent than previously estimated.
Our findings also highlight the differences between two exten-
sively used screening tests of global cognition, that is, MMSE and
MoCA. Our data indicate a ceiling effect with MMSE, as a large pro-
portion of stroke survivors with normal MMSE were classified as cog-
nitively impaired by MoCA (Figure 2A). Additionally, MoCA classified
higher proportions of stroke survivors across all levels of cognitive
impairment, including severe cognitive impairment (Figure 2B). Also,
a larger proportion of stroke survivors displayed impaired visuoexec-
utive functioning with MoCA. It should be acknowledged that in the
absence of a more comprehensive neuropsychological test battery, the
accuracy of MMSE and MoCA against a gold standard could not be
determined in our study. Moreover, we could not determine whether
the MoCA results differed more than the MMSE results when com-
paring stroke survivors and controls because MoCA had not been
assessed in the latter group. Even so, we could compare MMSE and
MoCA results within the group of long-­term stroke survivors and sim-
ilar comparisons among short-­term stroke survivors without inclusion
of control subjects have been reported, as well as studies comparing
the relative performance of MMSE and MoCA regarding different cog-
nitive domains, for example, visuoexecutive functioning.11,17,18,20,28
The studies reported that the relative performance of MMSE and
MoCA depends on the type/degree of cognitive impairment aimed to
be detected.11,12,18,28 Both tests have been shown to perform simi-
larly in detecting dementia/multidomain MCI, whereas MMSE has a
 |
6       DELAVARAN et al.

T A B L E   2   Cognitive assessment results

Stroke survivors Control subjects
of study participants
Mini-­Mental State Examination (MMSE) (n=127)a (n=118)b (n=354)
Cognitive impairment, n (%)
None (score 27-­30) 69 (54.3) 63 (53.4) 179 (50.6)
Mild (score 23-­26) 34 (26.8) 32 (27.1) 132 (37.3)
Severe (score 0-­22) 24 (18.9) 23 (19.5) 43 (12.1)
Montreal Cognitive Assessment (MoCA) (n=122)c
Cognitive impairment, n (%)
None (score 25-­30) 47 (38.5) –­
Mild (score 20-­24) 40 (32.8) –­
Severe (score 0-­19) 35 (28.7) –­
a
All 10-­year stroke survivors included.
b
All 10-­year stroke survivors aged ≥60 years and for whom controls were available.
c
All 10-­year stroke survivors who completed MoCA.

F I G U R E   2   Bubble plot illustrating the Mini-­Mental State Examination (MMSE) vs the Montreal Cognitive Assessment (MoCA) scores among
the 10-­year stroke survivors (n=122). Dark-­shaded bubbles represent stroke survivors with divergent classifications by MMSE and MoCA. (A)
The vertical and horizontal dashed lines illustrate the cutoffs for any cognitive impairment defined by MMSE score <27 and MoCA score <25,
respectively. (B) The vertical and horizontal dashed lines show the cutoffs for severe cognitive impairment defined by MMSE score <23 and
MoCA score <20, respectively

T A B L E   3   Odds for the presence of cognitive impairment as evaluated by Mini-­Mental State Examination (MMSE) among 118 10-­year stroke
survivors compared to 354 non-­stroke control subjects

Any cognitive impairment (MMSE score <27) Severe cognitive impairment (MMSE score <23)

OR 95% CI P-­value OR 95% CI P-­value

Logistic regression model

Simple (stroke only) 0.89 0.59-­1.36 .60 1.75 1.00-­3.05 .048
Multiple (stroke+education) 1.03 0.66-­1.60 .91 2.48  1.34-­4.59  .004 

relatively low sensitivity for single-­domain MCI and a distinct ceiling
effect.11,12,18 Furthermore, it has been reported that MoCA is superior
to MMSE in detection of visuoexecutive abnormalities, which contrib-
utes to more variation in overall MoCA scores compared to MMSE, and
higher overall sensitivity for vascular cognitive impairment related to
cerebrovascular disease.20,26 Taken together, our findings support the
previously reported differences in the relative performance of MMSE
and MoCA and suggest that MoCA may be more suitable than MMSE
to assess PSCI long-­term after stroke. However, from a clinical point
of view, the preferred cognitive assessment instrument also depends
on the purpose of testing, and particularly the level/degree of cogni-
tive impairment aimed to be detected. It should also be emphasized
DELAVARAN et al.
that both MMSE and MoCA are screening tests of global cognitive
function, and scorings on MMSE or MoCA are guidance tools support-
ing the clinician in the decision whether to perform further cognitive
evaluations, rather than providing a diagnosis.
In our study, the stroke survivors had a higher educational level
than the controls, possibly because of slightly different geographical
uptake areas. When adjusting for this, the odds of having severe cog-
nitive impairment defined as MMSE<23 were 2.5 times higher among
the stroke survivors as compared to the controls. This finding supports
previous reports of an increased risk of dementia long-­term after
8,9
stroke compared to non-­stroke persons. We further found that 10-­
year stroke survivors performed worse on the MMSE visuoconstruc-
tion subtest compared to controls, suggesting a higher prevalence of
visuoexecutive deficits. This is consistent with previous studies with
shorter-­term follow-­ups reporting that visuoexecutive abnormalities
are common among TIA/stroke patients.20 We also noted that the
MMSE word recall subtest scoring of the controls was surprisingly low
(Table S2). There was no clear explanation for this, and we believe this
probably represents a random finding. Altogether, clinical follow-­ups
and vigilance for cognitive impairment seem warranted also long-­term
after stroke.
4.1 | Methodological aspects
Strengths of this study include the prospective population-­based de-
29,30
sign, the description of non-­participating stroke survivors, the
cognitive assessment of stroke survivors with two different instru-
ments, and the comparison to non-­stroke controls. However, some
methodological aspects merit mentioning.
As prestroke cognitive function was not assessed, the degree to
which the measured PSCI was secondary to stroke vs to pre-­existing
impairment could not be established. Still, this does not invalidate the
overall PSCI measurement. Besides, patients with prestroke cognitive
impairment are less likely to survive 10 years post-­stroke.31 Also, the
cognitive function of the stroke survivors was not assessed at base-
line, and therefore, we could not describe temporal changes of their
cognitive function during the 10 years after stroke. On the other hand,
our comparison with matched non-­stroke individuals indicates that
the prevalence and characteristics of cognitive impairment in 10-­year
stroke survivors differ from a control group.
Stroke survivors not available for assessment or unable to perform
MMSE and MoCA (12%; Figure 1), as well as possible selective attri-
tion due to death before the 10-­year follow-­up, may have caused un-
derestimation of PSCI in our study.30-32 The estimated prevalence of
PSCI may also have been influenced by recurrent strokes. It has previ-
ously been reported that recurrent stroke is an independent predictor
of PSCI, with prevalence rates of PSD being around three times as high
after recurrent stroke compared with first-­ever stroke.6 However, this
effect is probably limited in our study due to the relatively low number
of recurrent strokes.
Our estimates of PSCI are based on cognitive screening tests and
not on formal diagnostic criteria or comprehensive neuropsychological
evaluations, which might have yielded different results. However, MMSE
|
      7
and MoCA have shown acceptable accuracy in detecting PSCI and are
widely used.11,12,18,28,33 Nevertheless, the cognitive testing performance
may have been confounded by frequent post-­stroke sequelae such as
neglect, apraxia, agnosia, or mild-­to-­moderate language impairments,
as these deficits were not assessed.34 In particular, such undetected
deficiencies in larger network functions may negatively have impacted
the performance on the visuoexecutive subtests with subsequent lower
scoring results. Also, age-­ and education-­corrected MMSE and MoCA
cutoffs derived from normative data could have been used instead of
predefined cutoffs,35,36 but attempts to increase sensitivity with this
method generally decrease specificity and vice versa.14 In addition, the
order of cognitive testing may have influenced the results as the two
tests were not presented in a counterbalanced order across study par-
ticipants. In particular, as MoCA was always performed after MMSE, fa-
tigue may have negatively impacted the MoCA scorings even though we
tried to reduce this effect by keeping assessment sessions brief.
We prespecified to deliberately omit vascular risk factors, possibly
associated with cognitive impairment,6 from our multivariate analy-
sis because we did not aim to examine specific predictors of PSCI.
Furthermore, previous studies have shown that depression is frequent
among long-­term stroke survivors.37,38 Therefore, the fact that we did
not formally screen and subsequently adjust for depression may have
influenced our results, possibly causing an overestimation of PSCI in
our study. However, a previous study reported that neither depres-
sion nor anxiety, as measured on the Hospital Anxiety and Depression
Scale, were significantly correlated with the overall scores on the
MMSE or MoCA short-­term after stroke.28 Finally, as discussed above,
the controls had not performed MoCA, and comparisons in MoCA
performance between the 10-­year stroke survivors and controls may
possibly have yielded other differences in cognitive function between
these two groups than those found by MMSE.
In conclusion, PSCI with executive dysfunction is prevalent among
10-­year stroke survivors. The odds of having severe cognitive impair-
ment are higher among 10-­years stroke survivors as compared to non-­
stroke control persons. Our findings indicate that the prevalence of
long-­term PSCI might have been underestimated previously and also
suggest that MoCA may be more suitable than MMSE to assess PSCI
long-­term after stroke.
AC KNOW L ED G EM ENTS
We thank Björn Hansen for assistance in preparing the figures.
CO NFL I C TS O F I NT ER ES T
None.
REFERENCES
1. Leys D, Hénon H, Mackowiak-Cordoliani MA, Pasquier F. Poststroke
dementia. Lancet Neurol. 2005;4:752–759.
2. Claesson L, Lindén T, Skoog I, Blomstrand C. Cognitive impairment
after stroke—impact on activities of daily living and costs of care
for elderly people. The Göteborg 70+ Stroke Study. Cerebrovasc Dis.
2005;19:102–109.
 |
8       DELAVARAN et al.
3. Brainin M, Tuomilehto J, Heiss W-D, et  al. Post-­stroke cognitive
­decline: an update and perspectives for clinical research. Eur J Neurol.
2015;22:229–238.
4. Feigin VL, Forouzanfar MH, Krishnamurthi R, et al. Global and regional
burden of stroke during 1990-­2010: findings from the Global Burden
of Disease Study 2010. Lancet. 2014;383:245–254.
5. Prince M, Wimo A, Guerchet M, Ali G-C, Wu Y-T, Prina M. World
Alzheimer Report 2015. The Global Impact of Dementia. London:
Alzheimer’s Disease International; 2015.
6. Pendlebury ST, Rothwell PM. Prevalence, incidence, and factors asso-
ciated with pre-­stroke and post-­stroke dementia: a systematic review
and meta-­analysis. Lancet Neurol. 2009;8:1006–1018.
7. Narasimhalu K, Ang S, de Silva DA, et al. Severity of CIND and MCI
predict incidence of dementia in an ischemic stroke cohort. Neurology.
2009;73:1866–1872.
8. Kokmen E, Whisnant JP, O’Fallon WM, Chu CP, Beard CM. Dementia
after ischemic stroke: a population-­based study in Rochester,
Minnesota (1960-­1984). Neurology. 1996;46:154–159.
9. Ivan CS, Seshadri S, Beiser A, et  al. Dementia after stroke: the
Framingham Study. Stroke. 2004;35:1264–1268.
10. Douiri A, Rudd AG, Wolfe CD. Prevalence of poststroke cogni-
tive impairment: South London Stroke Register 1995-­2010. Stroke.
2013;44:138–145.
11. Pendlebury ST, Mariz J, Bull L, Mehta Z, Rothwell PM. Impact of dif-
ferent operational definitions on mild cognitive impairment rate and
MMSE and MoCA performance in transient ischaemic attack and
stroke. Cerebrovasc Dis. 2013;36:355–362.
12. Lees R, Selvarajah J, Fenton C, et al. Test accuracy of cognitive screen-
ing tests for diagnosis of dementia and multidomain cognitive impair-
ment in stroke. Stroke. 2014;45:3008–3018.
13. Folstein MF, Folstein SE, McHugh PR. “Mini-­mental state”. A practical
method for grading the cognitive state of patients for the clinician. J
Psychiatr Res. 1975;12:189–198.
14. Tombaugh TN, McIntyre NJ. The mini-­mental state examination: a
comprehensive review. J Am Geriatr Soc. 1992;40:922–935.
15. Grut M, Fratiglioni L, Viitanen M, Winblad B. Accuracy of the Mini-­
Mental Status Examination as a screening test for dementia in a
Swedish elderly population. Acta Neurol Scand. 1993;87:312–317.
16. Nasreddine ZS, Phillips NA, Bédirian V, et al. The Montreal Cognitive
Assessment, MoCA: a brief screening tool for mild cognitive impair-
ment. J Am Geriatr Soc. 2005;53:695–699.
17. Pendlebury ST, Cuthbertson FC, Welch SJ, Mehta Z, Rothwell PM.
Underestimation of cognitive impairment by Mini-­Mental State
Examination versus the Montreal Cognitive Assessment in patients
with transient ischemic attack and stroke: a population-­based study.
Stroke. 2010;41:1290–1293.
18. Pendlebury ST, Mariz J, Bull L, Mehta Z, Rothwell PM. MoCA, ACE-­R,
and MMSE versus the National Institute of Neurological Disorders
and Stroke-­Canadian Stroke Network vascular cognitive impairment
harmonization standards neuropsychological battery after TIA and
stroke. Stroke. 2012;43:464–469.
19. Hachinski V, Iadecola C, Petersen RC, et  al. National Institute of
Neurological Disorders and Stroke-­Canadian Stroke Network
vascular cognitive impairment harmonization standards. Stroke.
2006;37:2220–2241.
20. Mai LM, Sposato LA, Rothwell PM, Hachinski V, Pendlebury ST. A
comparison between the MoCA and the MMSE visuoexecutive sub-­
tests in detecting abnormalities in TIA/stroke patients. Int J Stroke.
2016;11:420–424.
21. Hallström B, Jönsson AC, Nerbrand C, Petersen B, Norrving B,
Lindgren A. Lund Stroke Register: hospitalization pattern and yield of
different screening methods for first-­ever stroke. Acta Neurol Scand.
2007;115:49–54.
22. WHO-MONICA. The World Health Organization MONICA Project
(monitoring trends and determinants in cardiovascular disease): a
major international collaboration. WHO MONICA Project Principal
Investigators. J Clin Epidemiol. 1988;41:105–114.
23. Brott T, Adams HP, Olinger CP, et al. Measurements of acute cerebral
infarction: a clinical examination scale. Stroke. 1989;20:864–870.
24. Jönsson AC, Delavaran H, Iwarsson S, Ståhl A, Norrving B, Lindgren
A. Functional status and patient-­reported outcome 10 years after
stroke: the Lund Stroke Register. Stroke. 2014;45:1784–1790.
25. van Swieten JC, Koudstaal PJ, Visser MC, Schouten HJ, van Gijn J.
Interobserver agreement for the assessment of handicap in stroke pa-
tients. Stroke. 1988;19:604–607.
26. Webb AJ, Pendlebury ST, Li L, et  al. Validation of the Montreal
Cognitive Assessment versus mini-­mental state examination against
hypertension and hypertensive arteriopathy after transient ischemic
attack or minor stroke. Stroke. 2014;45:3337–3342.
27. Lagergren M, Fratiglioni L, Hallberg IR, et al. A longitudinal study inte-
grating population, care and social services data. The Swedish National
study on Aging and Care (SNAC). Aging Clin Exp Res. 2004;16:158–168.
28. Cumming TB, Churilov L, Linden T, Bernhardt J. Montreal Cognitive
Assessment and Mini-­Mental State Examination are both valid cogni-
tive tools in stroke. Acta Neurol Scand. 2013;128:122–129.
29. Wall KJ, Isaacs ML, Copland DA, Cumming TB. Assessing cogni-
tion after stroke. Who misses out? A systematic review. Int J Stroke.
2015;10:665–671.
30. Pendlebury ST, Klaus SP, Thomson RJ, Mehta Z, Wharton RM,
Rothwell PM. Methodological factors in determining risk of dementia
after transient ischemic attack and stroke: (III) Applicability of cogni-
tive tests. Stroke. 2015;46:3067–3073.
31. Pendlebury ST, Chen PJ, Bull L, Silver L, Mehta Z, Rothwell PM.
Methodological factors in determining rates of dementia in transient
ischemic attack and stroke: (I) impact of baseline selection bias. Stroke.
2015;46:641–646.
32. Pendlebury ST, Chen PJ, Welch SJ, et al. Methodological factors in de-
termining risk of dementia after transient ischemic attack and stroke:
(II) effect of attrition on follow-­up. Stroke. 2015;46:1494–1500.
33. Kjörk E, Blomstrand C, Carlsson G, Lundgren-Nilsson Å, Gustafsson
C. Daily life consequences, cognitive impairment, and fatigue after
transient ischemic attack. Acta Neurol Scand. 2016;133:103–110.
34. Demeyere N, Riddoch MJ, Slavkova ED, et  al. Domain-­specific
versus generalized cognitive screening in acute stroke. J Neurol.
2016;263:306–315.
35. Crum RM, Anthony JC, Bassett SS, Folstein MF. Population-­based
norms for the Mini-­Mental State Examination by age and educational
level. JAMA. 1993;269:2386–2391.
36. Rossetti HC, Lacritz LH, Cullum CM, Weiner MF. Normative data for
the Montreal Cognitive Assessment (MoCA) in a population-­based
sample. Neurology. 2011;77:1272–1275.
37. Crichton SL, Bray BD, McKevitt C, Rudd AG, Wolfe CD. Patient outcomes
up to 15 years after stroke: survival, disability, quality of life, cognition
and mental health. J Neurol Neurosurg Psychiatry. 2016;87:1091–1098.
38. Koivunen RJ, Harno H, Tatlisumak T, Putaala J. Depression, anxiety,
and cognitive functioning after intracerebral hemorrhage. Acta Neurol
Scand. 2015;132:179–184.
S U P P O RT I NG I NFO R M AT I O N
Additional Supporting Information may be found online in the support-
ing information tab for this article.
How to cite this article: Delavaran, H., Jönsson, A.-C., Lövkvist, H.,
Iwarsson, S., Elmståhl, S., Norrving, B. and Lindgren, A. (2016),
Cognitive function in stroke survivors: A 10-­year follow-­up study.
Acta Neurologica Scandinavica, 00: 1–8. doi: 10.1111/ane.12709