American Journal of Epidemiology
Copyright © 1997 by The Johns Hopkins University School of Hygiene and Public Health
All rights reserved
Vol. 145, No. 12
Printed In USA.
Blood Pressure and Performance on the Mini-Mental State Examination in
the Very Old
Cross-sectional and Longitudinal Data from the Kungsholmen Project
Zhenchao Guo, Laura Fratiglioni, Bengt Winblad, and Matti Viitanen
The authors examined the association of blood pressure with cognitive function as assessed by the
Mini-Mental State Examination (MMSE) in a community-based Swedish cohort of 1,736 people aged 75-101
years. Age, sex, education, antihypertensive medication use, heart disease, and stroke were considered as
covariates. Multiple linear regression analysis indicated that both systolic and diastolic blood pressure,
measured in 1987-1989, were positively and significantly related to baseline MMSE score; baseline systolic
pressure was also positively and significantly related to follow-up MMSE score, measured after an average
period of 40.5 months among subjects who were not taking antihypertensive medication at baseline. Furthermore, in the nontreated group, multiple logistic regression showed that individuals with a baseline systolic
pressure less than 130 mmHg had an odds ratio of 1.88 (p = 0.05) for follow-up cognitive impairment (MMSE
score <24) compared with those whose systolic pressure was 130-159 mmHg. An increased but not
statistically significant risk of cognitive impairment was associated with high blood pressure (systolic pressure
£180 mmHg or diastolic pressure s95 mmHg) only in persons taking antihypertensive medication at baseline.
Subjects with systolic pressure of 160-179 mmHg tended to be at lower risk of cognitive impairment. These
results may support the view that a certain blood pressure level, particularly a systolic pressure of at least 130
mmHg, is important to the maintenance of cognitive functioning in the very old. They also suggest that severe
hypertension that is not well controlled (systolic pressure >180 mmHg or diastolic pressure >95 mmHg) is still
a threat to cognitive function in this age group. However, the use of blood pressure measurements made at
a single visit and the relatively short follow-up period should be considered when interpreting these results.
Am J Epidemiol 1997;145:1106-13.
aged; antihypertensive agents; blood pressure; cardiovascular diseases; cognition
An issue of growing interest is the association between blood pressure and cognitive decline in the
absence of clinically recognized stroke. Many clinical
studies have shown that hypertensive subjects perform
more poorly than normotensive subjects in one or
several areas of cognitive tasks (1-11), but a few
studies have failed to find this association (12-14).
Several epidemiologic studies have tried to clarify the
issue by investigating the relation between blood pressure and cognitive performance in the general elderly
population. Two reports from the Framingham Study
observed an inverse relation (15, 16), leading to the
Received for publication April 15,1996, and in final form February
24, 1997.
Abbreviations: ICD-8, International Classification of Diseases,
Eighth Revision; MMSE, Mini-Mental State Examination; SD, standard deviation.
From the Stockholm Gerontology Research Center and the Department of Geriatric Medicine, Karollnska Institute, Stockholm,
Sweden.
Reprint requests to Dr. Zhenchao Guo, Stockholm Gerontology
Research Center, Box 6401, S-113 82 Stockholm, Sweden.
conclusion that elevated blood pressure may be related
to a modest decline in cognitive functioning (15).
Wallace et al. (17) found that diastolic (but not systolic) hypertension was significantly related to lower
performance on a memory test. In contrast, Scherr et
al. (18) reported that neither systolic nor diastolic
blood pressure was related to cognitive performance in
the elderly.
Few studies in this field have focused specifically
on the very old. When the sample from the Framingham Study was stratified by age in another report,
Fanner et al. (19) observed that the cognitive performance of people aged 75 years or older was better on
several selected tests if they had either isolated systolic
or diastolic hypertension, although the association was
no longer significant after adjustment for other variables. A recent report from the Honolulu-Asia Aging
Study found that systolic blood pressure less than 110
mmHg was cross-sectionally related to poor cognitive
function in men aged 74-93 years, although increased
midlife systolic pressure was a significant predictor of
1106
Blood Pressure and Cognitive Function
reduced cognitive function in late life in this cohort
(20). These results imply that the relation between
blood pressure and cognitive performance in the very
old may differ from that observed in other age groups.
In this study, we examined the relation of baseline
blood pressure to baseline and follow-up cognitive
function in a geographically defined population with a
baseline age of 75 years or more.
MATERIALS AND METHODS
Study population
The Kungsholmen Project is a longitudinal study of
aging and dementia (21). The population base included all inhabitants of the Kungsholmen district of
Stockholm, Sweden, who were aged 75 years or older
on October 1, 1987. Of the eligible subjects, 1,736 (74
percent) completed both blood pressure recording and
cognitive testing at baseline. Of 1,736 baseline participants, 1,022 (59 percent) individuals were retested
after an average follow-up period of 40.5 months
(standard deviation (SD), 7.0). The baseline survey
was completed in April 1989. Follow-up cognitive
testing began in November 1990 and ended in December 1992. The cognitive test was administered using
the same procedures at baseline and follow-up.
1107
Information on drug use was collected for the 2
weeks preceding the baseline interview. Participants
were asked about use of both prescription and nonprescription drugs, and drug containers or prescription
forms were inspected to verify the information provided. If medications were administered by caregivers
or health care personnel, these persons were asked
about the participant's drug use. Antihypertensive
drugs included all medicines potentially used for lowering blood pressure, such as diuretics, /3-blockers,
calcium channel antagonists, a-blockers, and angiotensin-converting enzyme inhibitors (AnatomicTherapeutic-Chemical (ATC) classification system
(27) codes C02, C03, and C07). For 10 percent of the
subjects, data on drug use and education were obtained
from a proxy informant (relative, caregiver, or other).
Most baseline participants (80 percent) were examined
at the research center.
Information on medical history for all subjects was
collected from the computerized inpatient register system, which was started in 1969 and covers all hospitals
in the Stockholm area. Coronary heart disease (International Classification of Diseases, Eighth Revision
(ICD-8) (28), codes 410-414), cardiac dysrhythmia
(ICD-8 code 427), heart failure and other heart diseases (ICD-8 codes 428, 390-405, 415-426, and
429), and stroke (ICD-8 codes 430-438) were included in the analyses as covariates.
Measurements
Cognitive function was assessed by means of the
Mini-Mental State Examination (MMSE) (22), which
had been directly translated into Swedish with no
major modifications. The MMSE is a screening test
for dementia, and it is also frequently used to evaluate
cognitive impairment in epidemiologic studies (2326). It measures several domains of cognitive function,
yielding a possible total score ranging from 0 (worst)
to 30 (best) points. Both serial subtraction of 7's and
backwards spelling of the word "konst" (the Swedish
word for art) are included in the Swedish version of
the MMSE, but only the higher score from these two
test domains was used for analysis.
Arterial blood pressure was measured with a mercury sphygmomanometer with the subject sitting after
having rested for 5 minutes. Systolic and diastolic
pressure were defined as Korotkoff phases 1 and 5,
respectively. If the first reading was abnormal, two
additional readings were taken, and the mean of the
second and third readings was used for analysis. Although the cognitive testing and blood pressure measurement were usually conducted by the same nurse at
baseline, neither nurses nor participants were aware
that blood pressure would subsequently be related to
cognitive performance.
Am J Epidemiol
Vol. 145, No. 12, 1997
Statistical analyses
A multiple linear regression model was used to
examine the association between blood pressure and
MMSE score, where MMSE score was the dependent
variable and blood pressure was the independent variable. Baseline and follow-up MMSE score were analyzed separately with baseline blood pressure. In analyses of the total population, systolic and diastolic
pressure were separately entered into the model with
all of the covariates (age, sex, education, antihypertensive medication use, heart disease, and stroke).
Analyses were also conducted in several subpopulations characterized by antihypertensive medication
use, age, sex, or cardiovascular disease. The regression
coefficient for blood pressure was expressed as the
mean change in MMSE score for every 10-mmHg
increase in blood pressure. The distribution of MMSE
scores was skewed toward high values. However, no
data transformations, including logarithmic transformation of MMSE scores, substantially improved the
linearity or strength of the association between blood
pressure and MMSE score. Thus, results from the
models based on the actual MMSE scores are presented here.
1108
Guoetal.
A multiple logistic regression model was used to
calculate odds ratios for cognitive impairment, defined
as an MMSE score less than 24.
Age was included as a continuous variable in all
models. Coronary heart disease, cardiac dysrhythmia,
and heart failure and other heart diseases were combined into one variable because their data overlapped
to a large extent. There were two people with less than
4 years of education in this population. A significant
difference in MMSE score was found only between
persons with < 8 years of education and those with > 8
years of education. Therefore, education was treated as
a dichotomous variable (<8 vs. s 8 years). As expected, results were the same regardless of whether
education was included in the models as a continuous
or a categorical variable. All analyses were conducted
with SPSS statistical software (SPSS, Inc., Chicago,
Illinois).
RESULTS
Table 1 shows numbers of subjects, mean MMSE
scores, and the prevalence of MMSE scores less than
24 according to various factors in the baseline popu-
TABLE 1. Numbers of subjects, mean MMSE* scores, and prevalence of MMSE scores less than 24 in the baseline study
population, according to blood pressure level and all covariates: The Kungsholmen Project, Stockholm, Sweden, 1987-1989
Total
population
Variable
%wtlh
MMSE
No use of
artJhypertenstve medication
%wtth
No.
Mean
MMSE
of
MMSE
SCOTS
subjects
score
<24
982
24.4
22.1
No.
of
subjects
Mean
MMSE
score
1,736
24.8
<24
20.0
75-79
80-84
85-101
669
549
518
26.4
25.3
22.4
10.9
17.1
34.9
394
310
278
26.4
24.8
21.2
Sex
Mala
Female
418
1,318
25.5
24.6
17.0
21.0
269
713
927
469
340
24.0
25.9
25.8
25.7
13.0
14.4
Coronary heart disease
No
Vfes
1,577
159
25.0
23.1
Cardiac dysrhythmia
No
Yss
1,559
177
Heart failure and other heart
diseases
No
\fes
Stroke
No
\fes
Total
Use of
anrJhypertenstve medication
%wth
No.
of
subjects
Mean
MMSE
score
754
25.4
17.4
11.7
19.7
39.6
275
239
240
26.4
26.0
23.8
9.8
13.8
29.6
25.4
24.0
16.7
24.1
149
605
25.6
25.4
17.4
17.4
508
268
206
23.4
25.4
25.4
27.4
16.8
16.0
419
201
134
24.6
26.6
26.3
23.6
8.0
11.9
18.9
31.4
936
46
24.5
21.2
21.6
32.6
641
113
25.7
23.9
15.0
31.0
25.1
23.0
18.7
32.2
927
55
24.6
21.0
21.4
34.5
632
122
25.7
23.9
14.7
31.1
1,597
139
25.1
21.5
18.3
40.3
944
38
24.7
16.8
20.7
57.9
653
101
25.8
23.2
14.9
33.7
1,591
145
25.2
215
18.1
41.4
909
73
24.8
19.0
19.7
52.1
682
72
25.6
23.5
16.0
30.6
198
738
493
307
21.3
25.0
25.4
26.0
41.9
20.1
15.8
12.7
130
435
256
161
20.8
24.8
24.6
25.8
42.3
20.0
20.7
13.7
68
303
237
146
2Z3
25.2
26.1
26.2
41.2
20.1
10.5
11.6
400
1,094
242
23.3
25.2
25.9
29.5
17.6
15.7
230
621
131
22.6
24.7
26.0
32.2
20.1
13.7
170
473
111
24.1
25.8
25.7
25.9
14.2
18.0
8C0T6
MMSE
SCOTS
<24
Age (years)
Education (years)
<8t
8-10
>10
Systolic blood pressure (mmHg)
<130
130-159
160-179
£180
Oiastolic blood pressure (mmHg)
<75
75-94
£95
* MMSE, Mini-Mental State Examination (22).
t Two persons had less than 4 years of education.
Am J Epidemiol
Vol. 145, No. 12, 1997
Blood Pressure and Cognitive Function
lation. People with low blood pressure (systolic pressure <130 or diastolic pressure <75 mmHg) had a
lower mean MMSE score and a higher prevalence of
MMSE scores under 24 than those with higher blood
pressure. The group of people taking antihypertensive
medication had a higher mean MMSE score and
higher proportions of subjects with heart disease (coronary heart disease, cardiac dysrhythmia, and heart
failure and other heart diseases) than the group of
persons who were not taking antihypertensive medication.
Table 2 lists the multiple linear regression coefficients for the relation between MMSE score and each
independent variable in the baseline population. In any
group, both systolic and diastolic blood pressure were
positively and significantly related to MMSE score.
For example, every 10-mmHg increase in systolic
pressure was associated with a 0.49-unit increase in
MMSE score in the group not taking antihypertensive
medication. Overall, systolic pressure in combination
with all of the covariates explained 20 percent of the
variance in MMSE scores, and systolic pressure alone
accounted for approximately 3 percent of the MMSE
score variance. Use of antihypertensive medication
was correlated with a 1.66-unit increase in MMSE
score, and it reduced in varying amounts the association of all other variables with MMSE score.
Blood pressure was inversely correlated with age in
the baseline population. Pearson correlation coefficients were —0.09 (p < 0.01) for systolic pressure and
—0.19 (p < 0.01) for diastolic pressure. For additional
control for the potential effect of age on the relation
between blood pressure and cognitive performance,
we stratified the baseline total population into three
groups according to age (table 3). The regression coefficient for systolic pressure, adjusted for all of the
1109
covariates, including age as a continuous variable, was
significant in all three age groups, while the significant
regression coefficient for diastolic pressure emerged
only in the oldest age group. Table 3 also lists the
regression coefficients for the relation between baseline blood pressure and baseline MMSE score in subpopulations characterized by sex or cardiovascular
disease. The positive association of systolic pressure
with MMSE score was consistently significant in all
subpopulations, but the regression coefficient for diastolic pressure was not significant in men.
A total of 1,022 baseline individuals participated in
follow-up cognitive testing. Of the 714 nonrespondents, 423 (59.2 percent) had died and 31 (4.3 percent)
had moved during the follow-up period, while 260
(36.4 percent) refused to participate. People aged ^ 8 5
years at baseline had higher mortality than those aged
<85 years (35.1 percent vs. 17.7 percent). Although
the follow-up and baseline populations had similar
mean MMSE scores (24.6 (SD, 5.6) vs. 24.8 (SD,
5.8)), the proportion with an MMSE score less than 24
was higher in the follow-up population than in the
baseline population (25.4 percent vs. 20.0 percent).
Table 4 gives the multiple linear regression coefficients for the relation between baseline blood pressure
and follow-up MMSE score. Baseline systolic pressure was positively related to follow-up MMSE score
in the total population and in persons not taking antihypertensive medication at baseline. The regression
coefficient for baseline diastolic pressure was not significant. Use of antihypertensive medication at baseline was associated with a higher follow-up MMSE
score (regression coefficient = 1.19, p = 0.0004).
Similar results were obtained in persons who had not
had cardiovascular disease before follow-up testing.
TABLE 2. Regression coefficients (p) for baseline MMSE* score in relation to baseline blood pressure and all covariates: The
Kungsholmen Project, Stockholm, Sweden, 1987-1
Total
population
variable
S£»
P
value
-0.29
-0.79
0.03
0.30
<0.0001
0.008
-1.57
-2.10
-2.97
0.41
0.45
1.66
0.25
0.33
0.46
0.06
0.11
0.26
<0.0001
<0.0001
<0.0001
<0.0001
0.0001
<0.0001
P
Age (years)
Female sex
<8 years of education (vs. £8
years)
Heart disease^
Stroke
Systolic blood pressure^
Diastolic Mood pressure§
Use of antihypertensive medication
No use of
anWiypertensVe medication
SE
P
value
-0.34
-1.18
0.04
0.41
<0.0001
0.004
-1.62
-3.79
-4.11
0.49
0.61
0.36
0.58
0.70
0.08
0.17
<0.0001
<0.0001
<0.0001
<0.0001
0.0003
P
Use of
arrtinypertenstve medication
S£
P
value
-0.22
-0.09
0.03
0.41
<0.0001
0.83
-1.54
-1.21
-1.60
0.34
0.30
0.33
0.36
0.55
0.07
0.15
<0.0001
0.007
0.004
<0.0001
0.04
P
* MMSE, Mini-Mental State Examination; SE, standard error.
f Systolic and diastoOc pressure were entered separately into a multiple linear regression model with all covariates; the coefficient of each
covanate was derived from the model which included all covariates and systolic blood pressure.
i Included coronary heart disease, cardiac dysrhythmia, and heart failure and other heart diseases.
§ Change per 10-mmHg increase.
Am J Epidemiol
Vol. 145, No. 12, 1997
1110
Guoetal.
TABLE 3. Regression coefficients (P) for baseline MMSE* score in relation to baseline blood pressure,
by age, sex, or cardiovascular disease: The Kungsholmen Project, Stockholm, Sweden, 1987-1989f
Systolic Wood pressure
(10-mmHg increase)
Dtastollc blood pressure
(10-mmHg increase)
P
P
SE*
P
value
P
SE
Age (years)
75-79
80-84
85-101
0.15
0.45
0.65
0.07
0.09
0.12
0.03
<0.0001
<0.0001
0.24
0.12
1.08
0.15
0.18
0.26
0.11
0.50
< 0.0001
Sex
Female
Male
0.43
0.32
0.07
0.10
<0.0001
0.002
0.52
0.24
0.14
0.20
0.0001
0.24
Cardiovascular disease
No
Mas*
0.33
0.63
0.06
0.14
<0.0001
<0.0001
0.38
0.69
0.12
0.28
0.001
0.01
value
* MMSE, Mini-Mental State Examination; SE, standard error.
t Systolic and diastolic pressure were entered separately into a multiple linear regression model with all
covanates except for the covariate by which the baseline total population was divided; age was included in all
models as a continuous variable.
t Included heart disease and stroke (n - 442).
TABLE 4. Regression coefficient* (p) for follow-up MMSE* score In relation to baseline blood
pressure: The Kungsholmen Project, Stockholm, Sweden, 1987-1902f
Blood
pressure
(10-mmHg Increase)
Total folow-tip
population
(n-1,022)
No use of
and hypertensive medication
at baselne (n =. 602)
Use of
andhypertenslve medcatfon
at baseline (n = 420)
P
SE»
P
value
P
SE
P
value
p
<*
JL
Systolic pressure
0.20
0.07
0.008
0.33
0.10
0.001
0.02
0.11
0.87
Diastolic pressure
0.12
0.15
0.44
0.28
0.21
0.18
-0.12
0.21
0.57
* MMSE, Mini-Mental State Examination; SE, standard error.
f Systolic and diastolic pressure were entered separately into a multiple linear regression model with all
covanates (age, sex, education, heart disease, stroke, and us© of antihypertensive medication).
Table 5 shows the odds ratios for follow-up cognitive impairment (MMSE score <24) from multiple
logistic regression analysis. There was a linear trend in
terms of a decreasing odds ratio for cognitive impairment according to increasing baseline systolic blood
pressure among persons who were not taking antihypertensive medication at baseline (trend test: p =
0.03). This trend was largely due to the elevated odds
ratio associated with a systolic pressure less than 130
mmHg (p = 0.05). High baseline blood pressure (systolic pressure £ 180 mmHg or diastolic pressure ^95
mmHg) was related to an elevated odds ratio in persons taking antihypertensive medication at baseline,
although the relation was not significant. Subjects with
systolic pressures of 160-179 mmHg tended to be at
lower risk of cognitive impairment.
To determine the risk of developing cognitive impairment, we excluded from the follow-up population
those who had had an MMSE score less than 24 at
baseline. There was no significant relation between
low or high blood pressure and the risk of developing
cognitive impairment among persons taking (n = 373)
or not taking antihypertensive medication at baseline.
However, antihypertensive medication use at baseline
was found to be related to a significantly decreased
risk of developing cognitive impairment (adjusted
odds ratio = 0.62, 95 percent confidence interval
0.42-0.93).
DISCUSSION
In this community-based population of very old
adults, we examined the relation between blood pressure and cognitive function as assessed by the MMSE.
In the baseline population, blood pressure (both systolic and diastolic) was strongly and positively related
to MMSE score. The relation emerged even in persons
taking antihypertensive medication and in those with
cardiovascular disease. Although age modified the relation somewhat, no opposite pattern was observed in
the youngest group, and the positive relation between
systolic pressure and MMSE score was significant in
all of the age groups we examined. In persons not
taking antihypertensive medication at baseline, baseAm J Epidemiol
Vol. 145, No. 12, 1997
Blood Pressure and Cognitive Function
1111
TABLE 5. Odd* ratio* relating baseline Mood pressure to follow-up cognitive impairment*: The
Kungsholmen Project, Stockholm, Sweden, 1987-1992f
Baseline
blood pressure
(mmHg)
Systolic pressure
Model 1§
Model 2
<130
130-159H
160-179
£180
Diastolic pressure
Model 1§
Model 2
<75
75-94H
No use of antlhypertsnslve
medlcaUon albaseSne
Use ot anflhypertenslve
medication at baseline
OR*
95%dt
OR
95% Cl
0.89
0.81-0.97
1.08
0.97-1.20
1.88
1.00
0.85
1.00-3.52
0.87
0.29-2.65
0.78
0.51-1.41
0.43-1.39
1.00
0.78
1.62
0.43-1.41
0.88-2.98
0.89
0.74-1.08
1.15
0.92-1.43
0.91
1.00
0.65
0.54-1.52
0.96
1.00
1.43
0.52-1.79
0.35-1.20
0.73-2.82
* Defined as a Mini-Mental State Examination (22) score less than 24.
t Systolic and diastolic pressure were analyzed separately in multiple logistic regression models adjusted for
age, sex, education, heart disease, and stroke.
$ OR, odds ratio; Cl, confidence interval.
§ Blood pressure was entered as a continuous variable (change per 10-mmHg increase).
H Reference category.
line systolic pressure was positively related to
follow-up MMSE score measured after an average
period of 40.5 months.
Compared with a medium level of systolic blood
pressure (130-159 mmHg), low systolic pressure
(<130 mmHg) was related to an elevated odds ratio
(p = 0.05) for follow-up cognitive impairment
(MMSE score <24) in persons who were not taking
antihypertensive medication at baseline, and systolic
pressures of 160-179 mmHg tended to be related to a
decreased odds ratio for follow-up cognitive impairment. However, systolic pressure 5:180 mmHg or
diastolic pressure ^95 mmHg was related to an increased odds ratio for follow-up cognitive impairment
(although the relation was not statistically significant)
in persons taking antihypertensive medication at baseline.
Our finding of a positive relation between blood
pressure and MMSE score above the age of 75 years is
in agreement with unpublished data from the Rotterdam Study, where similar results from linear regression analyses were obtained (29). A recent report from
the Honolulu-Asia Aging Study found that systolic
pressure less than 110 mmHg was cross-sectionally
related to poor cognitive function in men aged 74-93
years. These results are, to some degree, in accordance
with the clinical observation that Alzheimer's disease
patients have lower arterial blood pressures than nondemented individuals (30, 31). Our previous study also
observed that dementia patients had lower blood pressures in this cohort (32).
Am J Epidemiol
Vol. 145, No. 12, 1997
Based on the observation that low blood pressure
increases the short-term risk of mortality but predicts
better long-term survival, it has been suggested that
elderly persons with low blood pressure may be a
mixture of two subgroups: people with low blood
pressure due to multiple chronic diseases and people
with low blood pressure who are in good health (33).
This hypothesis may explain our finding of an attenuated association between low blood pressure and
follow-up cognitive impairment. Similarly, the relation of low blood pressure to poor cognitive function
has been attributed to the presence of some diseases
which may be responsible for both the blood pressure
reduction and cognitive decline (20). However, adjustment for hospitalization for heart disease and stroke
did not significantly modify the relation in this study.
Our results concerning mildly elevated systolic
blood pressure (160-179 mmHg), which was found to
be no longer harmful to cognitive function in the very
old, do not diminish the importance of the findings
that elevated midlife blood pressure may predict
poorer late-life cognitive function (15, 20, 34), since
our data also suggest that severe hypertension that is
not well controlled (systolic pressure ^180 mmHg or
diastolic pressure ^95 mmHg) is still harmful in this
age group. There is no doubt that hypertension should
be treated in middle-aged persons.
We found that antihypertensive medication use at
baseline was related to a 40 percent decreased risk of
developing cognitive impairment. The possible mechanism might be that antihypertensive medications re-
1112
Guoetal.
duce the risk of cognitive impairment by lowering the
incidence of clinically unrecognized cardiovascular
disease. However, it is also possible that some mildly
hypertensive persons who are free of severe complications, such as cerebral lesions, may have a lower risk
of cognitive impairment.
Several possible limitations of our study should be
considered. First, recording of blood pressure at a
single visit may diminish the accuracy of the data (35,
36). Such imprecision in blood pressure measurement would bias the relation towards the null. This
limitation may at least partially contribute to the
weak relation found between diastolic pressure and
follow-up MMSE score. However, it is possible that
systolic pressure has a stronger association with cognitive function than diastolic pressure (20). Particularly, such misclassification may reduce the power to
detect the difference in follow-up MMSE score between those with systolic pressure of 130-159 mmHg
and those with systolic pressure of 160-179 mmHg.
Second, the MMSE is a brief cognitive test which
may not detect subtle cognitive changes. However, the
MMSE is frequently used as a research instrument in
epidemiologic studies (24-26), and it has been well
validated as a screening test for dementia (23). In this
population, at the cutoff point of an MMSE score less
than 24, the estimated sensitivity for dementia was 87
percent and the specificity was 92 percent (37).
A third limitation is that the data on cardiovascular
disease were collected from the computerized inpatient register system, which would identify the more
severely ill and miss mild cases of illness. To determine the effect this might have had on our results, we
selected a sample of 624 persons from the baseline
population and conducted analyses using cardiovascular disease data from the interview with the subject or
proxy. The results pertaining to the association of
MMSE score with blood pressure and stroke were
similar to those derived from the entire baseline population using the data from the register system, but the
association with heart disease was markedly reduced
(regression coefficient = — 0.58, p = 0.38). This result
could reflect the inaccuracy of recall data in older
people; but more likely, in our opinion, it suggests that
mild heart disease is unrelated to cognitive function.
This conclusion was supported by further analysis
showing that use of cardiac glycosides, which probably reflects the presence of heart failure, was not
significantly related to baseline and follow-up MMSE
scores, and it did not modify the relation between
blood pressure and cognitive performance (data not
shown).
The fourth limitation of this study is that we did not
obtain detailed information on other potentially con-
founding factors, such as smoking, alcohol intake, and
occupation. However, there is no reason to believe that
smoking and alcohol use would substantially bias our
results, in light of the uncertain effect of smoking on
cognitive function (9, 25) and the low rate of alcohol
abuse in Sweden (38).
Finally, our analyses failed to show that baseline
low systolic blood pressure may significantly increase
the risk of developing cognitive impairment. Unfortunately, the limited sample size and the attrition rate
make conclusive statements impossible. We cannot
determine with certainty the direction of the observed
association of low systolic pressure with cognitive
impairment.
In conclusion, it seems that the relation between
blood pressure and cognitive function is more complicated in the very old than in other age groups. The
results of this study show that untreated systolic pressure is positively related to cognitive performance
above the age of 75 years, and they particularly emphasize the relation of low blood pressure to poor
cognitive function. Our data may support the view that
a certain level of blood pressure, particularly a systolic
pressure of at least 130 mmHg, is necessary to maintain cerebral perfusion and thereby to preserve cognitive ability (39), especially for those aged 75 years or
more. However, our data also suggest that severe
hypertension that is not well controlled (systolic pressure S:180 mmHg or diastolic pressure ^ 9 5 mmHg) is
still a threat to cognitive function in this age group.
Multidisciplinary studies are needed to clarify the
causal relation between blood pressure and cognitive
function, and cognitive function should be taken into
account in further evaluations of blood pressure's impact on mortality and other health outcomes in the
very old.
ACKNOWLEDGMENTS
This study was supported financially by the Swedish
Medical Research Council, the Swedish Council for Social
Research, the Swedish Municipal Pension Institute, the Torsten and Ragnar Soderbergs Foundation, and the Foundation for Medical Research (Stiftelsen Hjalp till Medicin
Forskning).
The authors thank all Kungsholmen Project workers for
collecting and managing data, and Phillippa Wills for her
assistance with manuscript preparation.
REFERENCES
1. Shapiro AP, Miller RE, King HE, et al. Behavioral consequences of mild hypertension. Hypertension 1982;4:355-60.
2. Madden DJ, Blumenthal JA. Slowing of memory-search perAm J Epidemiol
Vol. 145, No. 12, 1997
Blood Pressure and Cognitive Function
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
formance in men with mild hypertension. Health Psychol
1989;8:131-42.
Battersby C, Hartley K, Fletcher AE, et al. Cognitive function
in hypertension: a community based study. J Hum Hypertens
1993;7:117-23.
Waldstein SR, Ryan CM, Manuck SB, et al. Learning and
memory function in men with untreated blood pressure elevation. J Consult Clin Psychol 1991;59:513-7.
Schmidt R, Fazekas F, Offenbacher H, et al. Magnetic resonance imaging white matter lesions and cognitive impairment
in hypertensive individuals. Arch Neurol 1991;48:417-20.
van Swieten JC, Geyskes GG, Derix MM, et al. Hypertension
in the elderly is associated with white matter lesions and
cognitive decline. Ann Neurol 1991;30:825-30.
Kuusisto J, Koivisto K, Mykkanen L, et al. Essential hypertension and cognitive function: the role of hyperinsulinemia.
Hypertension 1993;22:771-9.
Elias MF, Robbins MA, Schultz NR Jr, et al. Is blood pressure
an important variable in research on aging and neuropsychological test performance? J Gerontol 1990;45:P128-35.
Colsher PL, Wallace RB. Epidemiologic considerations in
studies of cognitive function in the elderly: methodology and
nondementing acquired dysfunction. Epidemiol Rev 1991; 13:
1-27.
Starr JM, Whalley LJ. Senile hypertension and cognitive
impairment: an overview. J Hypertens Suppl 1992;10:S3142.
Schmidt R, Fazekas F, Koch M, et al. Magnetic resonance
imaging cerebral abnormalities and neuropsychologic test performance in elderly hypertensive subjects: a case-control
study. Arch Neurol 1995,52:905-10.
Desmond DW, Tatemichi TK, Paik M, et al. Risk factors for
cerebrovascular disease as correlates of cognitive function in
a stroke-free cohort. Arch Neurol 1993 ;50:162-6.
Bird AS, Blizard RA, Mann AH. Treating hypertension in the
older person: an evaluation of the association of blood pressure level and its reduction with cognitive performance. J Hypertens 1990;8:147-52.
Perez-Stable EJ, Coates TJ, Halliday R, et al. The effects of
mild diastolic hypertension on the results of tests of cognitive
function in adults 22 to 59 years of age. J Gen Intern Med
1992;7:19-25.
Elias MF, Wolf PA, D'Agostino RB, et al. Untreated blood
pressure level is inversely related to cognitive functioning:
The Framingham Study. Am J Epidemiol 1993;138:353-64.
Farmer ME, Kittner SJ, Abbott RD, et al. Longitudinally
measured blood pressure, antihypertensive medication use,
and cognitive performance: The Framingham Study. J Clin
Epidemiol 1990;43:475-80.
Wallace RB, Lemke JH, Morris MC, et al. Relationship of
free-recall memory to hypertension in the elderly: The Iowa
65+ Rural Health Study. J Chronic Dis 1985;38:475-81.
Scherr PA, Hebert LE, Smith LA, et al. Relation of blood
pressure to cognitive function in the elderly. Am J Epidemiol
1991:134:1303-15.
Farmer ME, White LR, Abbott RD, et al. Blood pressure and
cognitive performance: The Framingham Study. Am J Epidemiol 1987;126:1103-14.
Launer LJ, Masaki K, Petrovitch H, et al. The association
between midlife blood pressure levels and late-life cognitive
function: The Honolulu-Asia Aging Study. JAMA 1995;274:
Am J Epidemiol
Vol. 145, No. 12, 1997
1113
1846-51.
21. Fratiglioni L, Viitanen M, Backman L, et al. Occurrence of
dementia in advanced age: the study design of the Kungsholmen Project Neuroepidemiology 1992;ll(suppl l):29-36.
22. 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;12:189-98.
23. Tombaugh TN, Mclntyre NJ. The Mini-Mental State
Examination: a comprehensive review. J Am Geriatr Soc
1992;40:922-35.
24. Jacqmin H, Commenges D, Letenneur L, et al. Components of
drinking water and risk of cognitive impairment in the elderly.
Am J Epidemiol 1994;139:48-57.
25. Launer LJ, Feskens EJ, Kalmijn S, et al. Smoking, drinking,
and thinking: The Zutphen Elderly Study. Am J Epidemiol
1996;143:219-27.
26. Fraser GE, Singh PN, Bennett H. Variables associated with
cognitive function in elderly California Seventh-day Adventists. Am J Epidemiol 1996;143:1181-90.
27. Nordic Council on Medicines. Guidelines for ATC classification. Uppsala, Sweden: Nordic Council on Medicines, 1985.
(NLN publication no. 16).
28. World Health Organization. International classification of diseases. Manual of the international statistical classification of
diseases, injuries, and causes of death. Eighth Revision. Vol 1.
Geneva, Switzerland: World Health Organization, 1967.
29. Breteler MM. Cognitive decline in the elderly. (PhD dissertation). Rotterdam, The Netherlands: Department of Epidemiology and Biostatistics, Erasmus University Medical School,
1993.
30. Landin K, Blennow K, Wallin A, et al. Low blood pressure
and blood glucose levels in Alzheimer's disease: evidence for
a hypometabolic disorder. J Intern Med 1993;233:357-63.
31. Wang SJ, Liao KK, Fuh JL, et al. Cardiovascular autonomic
functions in Alzheimer's disease. Age Ageing 1994;23:
400-4.
32. Guo Z, Viitanen M, Fratiglioni L, et al. Low blood pressure
and dementia in elderly people: The Kungsholmen Project
BMJ 1996;312:805-8.
33. Glynn RJ, Field TS, Rosner B, et al. Evidence for a positive
linear relation between blood pressure and mortality in elderly
people. Lancet 1995;345:825-9.
34. Skoog I, Lernfelt B, Landahl S, et al. 15-year longitudinal
study of blood pressure and dementia. Lancet 1996;347:
1141-5.
35. Llabre MM, Ironson GH, Spitzer SB, et al. How many blood
pressure measurements are enough? An application of generalizability theory to the study of blood pressure reliability.
Psychophysiology 1988;25:97-106.
36. Pickering TG. Blood pressure measurement and detection of
hypertension. Lancet 1994;344:31-5.
37. Grut M, Fratiglioni L, Viitanen M, et al. Accuracy of the
Mini-Mental Status Examination as a screening test for dementia in a Swedish elderly population. Acta Neurol Scand
1993;87:312-17.
38. Spak F, Hallstrom T. Prevalence of female alcohol dependence and abuse in Sweden. Addiction 1995;90:1077-88.
39. Wang HS, Busse EW. Heart disease and brain impairment
among aged persons. In: Palmore E, ed. Reports from the
Duke Longitudinal Studies, 1970-1973. Durham, NC: Duke
University Press, 1974:160-7.