Sex and Age Differences in the Association of Blood Pressure and Hypertension with Cognitive Function in the Elderly: The Rancho Bernardo Study

Abstract

OBJECTIVES

This study examines sex and age differences in associations of systolic and diastolic blood pressure (SBP, DBP), pulse pressure and hypertension with cognitive function in a community-dwelling population.

DESIGN

Cross-sectional study.

SETTING

Research clinic visit in 1988-91.

PARTICIPANTS

Participants were 693 men and 1022 women aged 50-97

MEASUREMENTS

Blood pressure was measured and 12 cognitive function tests were administered.

RESULTS

Average age was 73.8±9.9 in men and 73.2±9.3 in women; 62.6% of men and 63.4% of women were hypertensive (SBP≥140 mmHg, DBP≥90 mmHg, or antihypertensive medication use). Each 5-unit increment in SBP, DBP, or pulse pressure and categorical hypertension was associated with significantly increased odds of poor verbal fluency performance in men and poor Trails B performance in women, with strongest associations for hypertension (OR=1.97, CI:1.01,3.85 in men; OR=1.51, CI:1.01,2.26 in women). After age stratification, associations remained statistically significant in younger (<80 years) but not older (≥80 years) participants.

CONCLUSION

Blood pressure as a continuous or categorical variable was associated with poor performance on cognitive function tests, but domains varied by sex and associations were found only in those younger than 80 years. The absent associations in those aged 80 years and older could support the hypothesis that increased blood flow is required to maintain cerebral perfusion with advancing age, or could reflect a survivor effect.

Introduction

Cognitive impairment and dementia are becoming increasingly prevalent with the aging of the population and carry a huge personal and economic burden (1-3). Therefore, identification of potentially modifiable risk factors for cognitive decline is important. One of the most plausible, potentially modifiable risk factors for cognitive decline is blood pressure because both elevated blood pressure and cognitive impairment are common in old age (2-5) and hypertension is the major risk factor for stroke (6-13). Small, undiagnosed strokes could themselves impair cognitive function, or hasten progression of cognitive impairment due to Alzheimer’s disease (13).

Previous studies examining the association of blood pressure with cognitive function have yielded contradictory results. Some studies reported that higher blood pressure (usually systolic blood pressure) and hypertension were associated with poorer cognitive function (14-21), while others reported a deleterious association with low blood pressure (20, 22-27), failed to find any association (28-30), or reported U- or J-shaped associations (23, 31, 32). Additionally, some age-stratified studies reported inverse associations between blood pressure and cognitive performance only among those older than 60 years of age (33) or only among the middle-aged (20).

Most previous studies of the association of blood pressure and cognition either did not include women or did not report data separately by sex. Although it has been reported that women are more likely to have high blood pressure than men (34-36), a meta-analysis suggests this is not true (37). Because of sex differences in performance on cognitive function tests (generally women perform better on verbal ability tasks and men perform better on spatial ability tests (38-44)), results of studies of the blood pressure-cognitive function association may vary by domain tested. Most previous studies included or reported only one or few cognitive function tests, or combined multiple tests into one composite measure. Yet, different tests identify specific domains of cognitive function that may be differentially associated with blood pressure.

The purpose of this study was to examine sex and age differences in the cross-sectional associations of systolic and diastolic blood pressure, pulse pressure, and categorically defined hypertension with different domains of cognitive function in community-dwelling older men and women.

Methods

Participants

In 1972-1974, 6339 individuals representing 82% of adults aged ≥30 living in the Southern California community of Rancho Bernardo were enrolled in a study of heart disease risk factors (45). Participants were predominantly Caucasian, relatively well-educated, and middle-class. In 1988-91, 82% of community-dwelling survivors (N=2,040) participated in a follow-up research clinic visit when cognitive function was assessed for the first time, blood pressure was measured, and information on behaviors was obtained. The 1715 individuals (693 men, 1022 women) aged 50 and older with concurrent measurement of blood pressure and administration of cognitive function tests are the focus of this report. All participants were ambulatory and gave written informed consent prior to participation; this study was approved by the Human Subjects Protection Program of the University of California, San Diego.

Procedures

In 1988-91, a self-administered questionnaire obtained information on cigarette smoking history (never/past/current), alcohol consumption (≥3 times/week; no/yes), and physical activity (exercise ≥3 times/week; no/yes). These self-reported behaviors were indirectly validated by the strong positive correlations of smoking with poor pulmonary function (46), alcohol use with higher serum aspartate aminotransferase (47) and high-density lipoprotein-cholesterol (48), and physical activity with pulse rate and high-density lipoprotein-cholesterol (49, 50). Current medication use, including anti-hypertensive medications, and estrogen therapy (in women), was validated by a nurse who examined containers and prescriptions brought to the clinic for that purpose.

Height (meters), weight (kilograms), and waist and hip circumferences (centimeters) were measured with the participants wearing light clothing without shoes. Body mass index (BMI; weight/height²) and waist-hip ratio (WHR; waist circumference/hip circumference X100) were used to estimate obesity and central adiposity, respectively.

After participants were seated quietly for 5 minutes, two blood pressure readings were taken at least 30 seconds apart by a nurse certified in the Hypertension Detection Follow-up Protocol (51), using a regularly calibrated standard mercury sphygmomanometer.

Educational level was queried and twelve standardized tests of diverse aspects of cognitive function were administered by a trained interviewer. These tests were selected with help from the UCSD Alzheimer’s Disease Research Center (ADRC) based on demonstrated reliability and validity (52, 53). Higher scores indicate better cognitive function except where noted. Cognitive function tests were:

The Buschke-Fuld Selective Reminding Test (54) assesses verbal episodic memory. Ten unrelated words are read to participants at a rate of one every 2 seconds. Immediately after, participants are asked to recall the entire list. They are reminded of any words they missed and asked to recall the entire list again. This procedure is followed for six trials. We analyzed three measures from this test, including the number of words on each trial in long term storage (a word is considered to have entered long term storage when it is recalled on two consecutive trials); the number in short term storage (words that are not consecutively recalled) and the total number of words in long and short term storage across the six trials (total recall, maximum=60). Higher scores on short-term memory indicate poorer performance because they reflect words not successfully encoded into long-term storage.

The Heaton Visual Reproduction Test (55), adapted from the Wechsler memory Scale (56), assesses memory for geometric forms. Three stimuli of increasing complexity are presented one at a time, for 10 seconds each. Participants are asked to reproduce the figures immediately to assess short-term memory and after 30 minutes of unrelated testing to assess long-term memory for geometric forms. Afterwards, participants are asked to copy the stimulus figures, which assess visuospatial impairments. Thus, three scores are obtained: immediate recall, delayed recall and copying.

The Mini-Mental State Examination (MMSE) (57, 58) a test of global function, assesses orientation, registration, attention, calculation, language, and recall. Total MMSE scores range from 0-30; persons with dementia usually score <24. Two MMSE items were analyzed separately; counting backward from 100 by sevens (Serial 7’s), which assesses calculation, and spelling the word “world” backwards (World backwards), both of which are measures of executive function. Maximum score is 5 for each.

Two items from the information-memory-concentration test of Blessed et al. (53) assess concentration by having the participant name the months of the year backward, and assess memory by asking participants to recall a five-part name and address following a 10-minute delay. Two points were given for correctly naming the months of the year backward and one point for each part of the name/address recalled correctly. Maximum score is 7.

The Trail-making Test, part B (Trails B), from the Halstead-Reitan Neuropsychological Test Battery (59), is a test of executive function that assesses visuomotor tracking and mental flexibility. Participants scan a page containing letters and numbers within circles and are asked to connect numbers and letters in ascending order, alternating between numbers and letters (e.g., 1 to A to 2 to B to 3 to C and so on). A maximum of 300 seconds is allowed; performance is rated by time required to finish the test; higher scores indicate poorer performance.

Category Fluency (60) assesses semantic fluency and verbal memory the participant is asked to name as many animals as possible in 1 minute. The score is the number of animals correctly named; repetitions, variants (e.g., dogs after producing dog), and intrusions (e.g., apple) are not counted.

Statistical analysis

Means of the two systolic (SBP) and two diastolic (DBP) blood pressure measurements were used for analyses of blood pressure as a continuous variable. Pulse pressure (PP) was calculated as SBP – DBP. Categorical hypertension was defined as SBP≥140 mmHg or DBP≥90 mmHg, or current use of antihypertensive medication. Cut points represent target blood pressure at the time data were obtained.

Because of previously reported sex differences in cognitive function scores in this cohort [38], all analyses were sex-specific. Education was dichotomized as no college versus some college or more. Means and distributions for continuous variables and rates for categorical variables were calculated. Comparisons of age between those with and without hypertension were performed with independent t-tests. Age-adjusted comparisons of characteristics and cognitive function test scores by hypertension status were performed with analysis of covariance for continuous variables and Mantel-Haenzel chi-square tests for categorical variables. Regression analyses examined associations of SBP, DBP, PP and categorical hypertension with scores on each cognitive function test after adjustment for age, education, estrogen use (in women), and antihypertensive medication use. Models evaluated linear and quadratic components for blood pressure variables: no quadratic components were statistically significant; results are shown only for linear components.

Cognitive function scores were also analyzed as categorical outcomes using cutoffs indicative of poor performance recommended by the UCSD ADRC. Cutoffs were available for five tests: Buschke Selective Reminding Test long-term memory (≤13), Heaton Visual Reproduction Test immediate recall (≤7), MMSE (≤24), Trails B (≥132) and Category Fluency (≤12). Because <1.5% scored below the Buschke long-term recall cutoff and ≤3.2% scored below the Heaton Visual Reproduction Test cutoff, results are presented only for the MMSE, Trails B, and Category Fluency. Logistic regression examined associations of each 5-unit change in SBP, DBP and PP, and presence of hypertension with risk of poor performance on each cognitive function test after adjustment for age, education, BMI, antihypertensive medication use, and, estrogen use in women. It was decided a priori to repeat analyses after stratification by age (<80 vs. ≥80 years) and antihypertensive medication use.

To examine participation bias, independent t-tests compared SBP and DBP from the 1984-87 research clinic visit between those who did and did not attend the 1988-91 clinic visit.

Statistical analyses were performed with SAS (version 8.1, SAS Institute, Cary, NC); all tests were two-tailed with p-value≤0.05 considered significant.

Results

In 1988-91, participants’ age ranged from 50-97 years (mean=73.8±9.9, median=74.3 for men, mean=73.2±9.3, median=73.1 for women). Table 1 shows the sex-specific distribution of characteristics. Based on categorically defined hypertension, (SBP≥140, DBP≥90, and/or use of antihypertensive medication), 62.6% of men and 63.4% of women were hypertensive. As previously reported, those with treated hypertension had higher blood pressure than those with untreated hypertension, reflecting the prescription of antihypertensive medications for those with higher blood pressure (61). Table 1 also shows mean scores on each cognitive function test for men and women, and rates of poor performance on tests with an available cutoff.

Table 1.

Baseline (1988-91) Characteristics for Men and Women

	Men (n=693)		Women (n=1022)
	Mean	SD	Mean	SD
Age (years)	73.8	9.9	73.2	9.3
Waist circumference (cm)	94.5	9.1	79.1	10.2
Waist-hip ratio	0.92	.05	0.79	.07
Body mass index (kg/m2)	26.0	3.4	24.5	4.0
Systolic BP (mmHg)	135.3	19.7	135.9	20.9
Diastolic BP (mmHg)	77.5	9.4	74.7	8.9
Pulse Pressure	58.3	17.2	61.3	17.9
Mini-Mental Status Exam	26.8	2.9	27.3	2.1
Trails B (seconds)	127.0	63.8	138.3	67.2
Verbal Fluency	18.5	5.4	17.5	4.8
Buschke Total Correct	35.2	9.7	40.0	9.1
Buschke Long Term Memory	27.5	12.9	34.2	12.3
Buschke Short Term Memory	7.7	4.6	5.9	4.5
Heaton Immediate Recall	10.2	3.9	9.4	3.6
Heaton Delayed Recall	7.8	4.7	6.7	4.0
Heaton Copying	15.2	2.1	15.2	2.1
Serial 7s	4.4	1.1	4.0	1.3
World Backwards	4.7	0.9	4.9	0.6
Blessed	6.0	1.5	6.1	1.3
	%		%
Hypertension† (%yes)	62.6		63.4
Blood pressure medications (%yes)	41.0		43.8
Some college or more (% yes)	77.3		61.8
Alcohol (≥3 drinks per week)	54.3		39.4
Ever smoked (% yes)	68.4		49.2
Current ERT use (% yes)	---		37.2
Exercise (≥3x per week)	38.7		30.3
Poor Buschke Long Term (% ≤13)	1.4		0.2
Poor MMSE (% ≤24)	11.4		6.0
Poor Heaton Immediate (% ≤7)	3.2		1.8
Poor Trails B (% ≥132)	34.2		41.4
Poor Verbal Fluency (% ≤12)	12.7		14.6

BP, blood pressure; ERT, estrogen replacement therapy; MMSE, mini-mental status exam;

^†Hypertension= SBP≥140 or DBP ≥90 or use of antihypertensive medications

Sex-specific comparisons of age and age-adjusted characteristics by hypertension status are shown in Table 2. Men and women with hypertension were significantly older (Ps<0.001), and had larger waist circumferences (Ps<0.001) and BMI (P=0.002 and P<0.001, respectively) than normotensive participants. There were no significant differences by hypertension status in education, alcohol use, smoking, exercise, or, in women, current estrogen use.

Table 2.

Age and Age-Adjusted Comparisons^§ of Characteristics by Hypertension Status 1988-91

	MEN			Women
	Normal	HTN†	P	Normal	HTN†	P
	(n=259)	(n=434)		(n=374)	(n=648)
Age (Years)	70.1	75.1	<0.001	69.8	75.2	<0.001
Waist Circumference (cm)	92.6	95.6	<0.001	76.8	80.4	<0.001
Body Mass Index (kg/m2)	25.4	26.3	0.002	23.8	25.0	<0.001
Some college or more (%)	52.2	52.7	0.906	28.1	26.6	0.634
Alcohol ≥3 drinks/week (%)	51.2	56.0	0.238	43.6	37.0	0.431
Ever Smoked (% yes)	66.7	69.4	0.486	46.2	51.0	0.158
Exercise ≥3x week (% yes)	41.9	36.7	0.190	33.9	28.3	0.061
Current ERT use	--	--	--	34.1	39.0	0.131

^§Independent t-tests used to compare age; analysis of covariance used to adjust for age;

^†HTN = hypertension = SBP≥140 or DBP ≥90 or use of antihypertensive medications

Age-adjusted comparisons of cognitive function by hypertension status showed women with hypertension had significantly lower scores than normotensive women on World Backwards (means=4.8 vs. 4.9, respectively, p=0.036) and Trails B (means=142.2 vs. 131.6, respectively, P=0.008); 45.0% of hypertensive women compared to 36.9% of normotensive women, scored above the Trails B categorical cut point of 132, indicating poor performance (P=0.007). Hypertensive women did not differ from normotensive women on other cognitive function tests. Analyses among men showed no significant differences on any cognitive function test by hypertension status.

Results of sex-specific models examining associations of SBP, DBP, PP, and hypertension with scores on each cognitive function test after adjustment for covariates are shown in Table 3. Women with categorical hypertension had poorer scores than normotensive women on World Backwards (β=-0.111, P=0.02) and Trails B (β=9.073, P=0.02). Hypertension was not associated with scores on the other cognitive function tests in women or with any cognitive function scores in men. There were no significant associations of SBP, DBP or PP with any cognitive function measures in either sex. No quadratic components in any models were significant indicating there was no U- or J-shaped association between blood pressure and cognitive function (data not shown).

Table 3.

Cross-sectional Multivariate Adjusted^A Associations of Blood Pressure and Cognitive Function

		SBP	DBP	Pulse-P	HTN
Men		β	β	β	β
	Buschke
	Total	-0.019	-0.081	0.034	-0.649
	Long term memory	-0.047	-0.076	-0.008	-1.110
	Short term memory	0.101	0.048	0.102	0.609
	Heaton Visual Recall
	Immediate	-0.015	-0.006	-0.017	-0.349
	Delayed	-0.017	-0.007	-0.029	-0.452
	Copying	-0.002	-0.003	-0.000	0.099
	MMSE
	Total	0.000	-0.042	0.031	0.406
	Serial 7s	-0.062	-0.032	-0.061	-0.256
	World backward	-0.031	0.028	-0.062	-0.046
	Blessed	0.000	-0.041	0.027	-0.107
	Trails B	0.024	0.037	0.006	2.462
	Verbal Fluency	-0.055	- 0.053	-0.037	-0.314
Women
	Buschke
	Total	-0.003	-0.044	0.025	-0.517
	Long term memory	-0.007	-0.005	0.022	-0.772
	Short term memory	-0.003	0.028	-0.062	-0.046
	Heaton Visual Recall
	Immediate	-0.006	-0.015	0.002	0.532
	Delayed	-0.020	-0.022	-0.012	0.555
	Copying	-0.007	-0.004	-0.007	0.136
	MMSE
	Total	-0.001	-0.014	0.006	0.139
	Serial 7s	-0.043	-0.002	0.058	-0.103
	World Backwards	0.014	-0.024	0.035	-0.111*
	Blessed	0.000	-0.041	0.027	-0.107
	Trails B	0.047	0.028	0.045	9.073**
	Verbal Fluency	-0.017	-0.025	-0.006	-0.008

DBP, diastolic blood pressure; HTN, hypertension; MMSE, mini-mental status exam; Pulse p, pulse pressure; SBP, systolic blood pressure;

^AResults of regression analyses adjusted for age, education, body mass index, current estrogen use (women only) and blood pressure medication use;

^Bnon-standardized beta; high blood pressure: SBP≥140 or DBP ≥90 or medication use;

^*P=0.0241,

^**P=0.0237

Age and education-adjusted cognitive function scores were also compared after dividing participants into four groups based on both hypertension status and antihypertensive medication use: normotensive and not using medication, controlled hypertensive (i.e., SBP and DBP measured at the clinic visit was within normal limits, but they were on antihypertensive medication), untreated hypertensive (not using medication), and treated but uncontrolled hypertensive. Compared to normotensive participants, men with controlled hypertension had higher MMSE scores (means=26.5 vs. 27.3 respectively, P=0.01) and women with uncontrolled hypertension had poorer scores (shown by a higher value) on Trails B (means=131.5 vs. 145.0, P=0.02). There were no other differences in cognitive function between groups based on hypertension status and antihypertensive medication use in either sex (data not shown).

Significant associations emerged in analyses of categorically defined poor performance on cognitive function tests adjusted for age and other covariates (Table 4). In men, a 5-unit change in any blood pressure variable was associated with significantly increased odds of poor performance on verbal fluency (SBP: OR=1.10, 95% Confidence Interval (CI):1.03,1.16; SBP: OR=1.15, CI:1.01,1.30; PP: OR=1.09, CI:1.02,1.17); those with hypertension had twice the odds of poor performance than those without hypertension (OR=1.97, CI:1.01,3.85). In women, each 5-unit increase in SBP and PP was associated with increased odds of poor performance on Trails B (OR=1.04, CI:1.01,1.08 for SBP; OR=1.05, CI:1.00,1.10 for PP), and hypertension was associated with over 50% increased odds of poor performance on Trails B (OR=1.51, CI:1.01,2.26).

Table 4.

Adjusted Odds^a of Categorically Defined Poor Performance by Blood Pressure and Hypertension Status for Men and Women Before and After Stratification for Age

		SBPb		DBPb		Pulse Pressureb		Hypertension
Men		OR	CI	OR	CI	OR	CI	OR	CI
MMSE	≤24 (% yes) All	0.99	0.93-1.06	1.01	0.88-1.15	0.98	0.91-1.06	1.08	0.54-2.14
	<80 years	1.09	0.96-1.24	0.99	0.77-1.28	1.13	0.97-1.32	1.56	0.46-5.25
	≥80 years	0.99	0.92-1.06	1.01	0.86-1.18	0.98	0.90-1.07	0.85	0.37-1.96
Trails B	≥132 (% yes) All	1.03	0.98-1.08	1.05	0.95-1.16	1.02	0.96-1.08	1.34	0.80-2.24
	<80 years	1.06	0.99-1.13	0.99	0.87-1.13	1.09*	1.01-1.18	1.65	0.89-3.04
	≥80 years	1.06	0.99-1.14	1.06	0.91-1.23	1.07	0.98-1.16	1.31	0.55-3.12
Verbal fluency	≤12 (% yes) All	1.10**	1.03-1.16	1.15*	1.01-1.30	1.09*	1.02-1.18	1.97*	1.01-3.85
	<80 yearsc	1.27**	1.12-1.43	1.40**	1.10-1.79	1.25**	1.07-1.45	2.16	0.75-6.22
	≥80 years	1.06	0.99-1.14	1.12	0.98-1.28	1.07	0.99-1.16	1.92	0.81-4.54
Women
MMSE	≤24 (% yes) All	1.01	0.94-1.08	1.08	0.93-1.25	0.99	0.91-1.07	0.98	0.48-2.01
	<80 years	1.07	0.96-1.18	0.97	0.75-1.26	1.11	0.98-1.26	1.92	0.51-7.26
	≥80 years	1.01	0.93-1.09	1.14	0.95-1.37	0.97	0.88-1.07	0.88	0.37-2.07
Trails B	≥132 (% yes) All	1.04*	1.01-1.08	1.05	0.97-1.13	1.05*	1.02-1.10	1.51*	1.01-2.26
	<80 years	1.13**	1.09-1.18	1.04	0.95-1.14	1.19**	1.13-1.25	2.75**	1.77-4.27
	≥80 years	0.99	0.92-1.05	1.00	0.86-1.16	0.98	0.91-1.06	1.36	0.63-2.94
Verbal fluency	≤12 (% yes) All	1.01	0.97-1.06	1.01	0.91-1.12	1.02	0.96-1.08	0.92	0.55-1.56
	<80 years	1.08**	1.02-1.14	0.98	0.85-1.13	1.12**	1.05-1.20	1.33	0.67-2.66
	≥80 years	1.02	0.96-1.09	1.01	0.87-1.16	1.03	0.95-1.11	1.27	0.58-2.77

DBP, diastolic blood pressure; MMSE, mini-mental status exam; SBP, systolic blood pressure;

^*P≤0.05,

^**P≤0.01;

^aResults of logistic regression analysis adjusted for age, education, body mass index, current estrogen use (women only) and blood pressure medication use;

^bOdds ratio calculated per 5 units of change in systolic and diastolic blood pressures and pulse pressure

Analyses stratified by age (<80 vs. ≥80+ years) showed that odds of poor performance on verbal fluency in men and Trails B in women increased significantly with higher blood pressure and/or categorical hypertension only for those younger than 80 years of age. Additionally, among women younger than 80 years, each 5-unit increase in SBP and PP was associated with significantly increased risk of poor performance on verbal fluency (OR=1.08 CI:1.02,1.14 for SBP; OR=1.12 CI:1.05,1.20 for PP). Among men and women aged 80 and older, blood pressure and hypertension were not associated with odds of poor performance on any cognitive function test (see Table 4). Comparisons of blood pressure showed that in both sexes, participants aged ≥80 years had significantly higher SBP compared to those <80 years (mean SBP=143.8 for ages ≥80 and =131.8 for ages <80 in men; mean SBP=143.2 for ages ≥80 and =133.2 for ages <80 in women, P’s <0.001). There were no differences in DBP by age 80 years in men or women (P’s >0.10).

Sensitivity analyses for participation bias showed SBP was lower among those who came to both the 1984-87 and 1988-91 clinic visits (means=132.7 vs. 142.0, respectively, for both sexes combined, P<0.001) than those who did not return.

Discussion

In this study of older adults, higher SBP, DBP, PP, and categorical hypertension were associated with significantly increased odds of poorer performance on verbal fluency in men, and higher SBP, PP and hypertension were significantly associated with categorically defined poorer performance on Trails B, a test of executive function, in women. These associations were independent of education, BMI, blood pressure medication use, and estrogen therapy in women. However, for both sexes, increased risk was confined to those younger than 80 years; among those older than 80 years, there was no association of blood pressure or hypertension with performance on cognitive function tests despite higher SBP among the older individuals. To our knowledge, this is the first study to investigate the sex-specific association of blood pressure and hypertension with multiple domains of cognitive function before and after stratification by age.

Women with hypertension had significantly poorer scores than normotensive women on Trails B and World Backwards as continuous variables, but there were no other significant associations of SBP, DBP, and PP with continuous cognitive function scores in either sex, suggesting that blood pressure—cognitive function associations are small and only observable when large gradations in performance are examined.

These results are in accord with other studies reporting associations of hypertension and blood pressure with poorer cognitive function (14-21). For example, Kuo et al. (19) reported that each 10mmHg increase in supine SBP was associated with a 2.31 increased odds (CI=1.14-4.66) of poor Trails B performance in a combined sample of men and women aged 65-85. In that study, risk for cognitive impairment increased steeply when supine SBP was greater than 135mmHg (19), similar to the categorical definition of hypertension used in this study. Knecht et al. (20) reported an inverse linear association between SBP and a composite of eight cognitive function tests in 377 community-dwelling men and women; these data were not described separately by sex. Obisesan et al. (33) in 6163 men and women aged ≥60 from NHANES III, also reported that hypertension was associated with poorer MMSE performance but sex-specific results were not reported.

Results of our study contrast with Keary et al. (62), who reported that greater blood pressure variability was associated with better performance on the Dementia Rating Scale among 97 older men and women (average age 70) with cardiovascular disease. Likewise, Guo and colleagues (22) reported better MMSE scores with increasing blood pressure among 1,736 participants aged 75-101 years in the Kungsholmen Project, whereas we found no associations in men and women aged 80 and older in the Rancho Bernardo cohort. Furthermore, unlike our study, age and sex-specific data were not reported by Guo et al. (22), nor were specific separate verbal and spatial ability scores.

In our study we examined sex-specific associations of blood pressure with cognitive function in several ways; blood pressure was used as continuous as well as categorical variables (hypertension). Cognitive function scores were examined as both continuous and categorical outcome variables. In this cohort, continuous blood pressure variables were generally not significantly associated with cognitive function scores nor were there significant nonlinear trends, in contrast with reports of cognitive function decrements at very high and very low blood pressures (a U-shaped or J-shaped distribution) (23, 24). Similar to a previous study (31), we found that high blood pressure was associated with poorer cognitive function and impairment only in younger individuals. Men and women ≥80 years had significantly higher SBP than those younger than 80, but there was no detrimental effect of high blood pressure. It is biologically plausible that associations of blood pressure with cognitive function vary by age. Due to stiffening of the arteries with age, older individuals may require higher blood pressure for brain perfusion (22, see reviews, 63, 64). Our results are in accord with recent guidelines suggesting that for adults older than 60 years, SBP <150 mmHg is adequate and does not need to be managed (65). This guideline also recognizes that treating to target below 140 mmHg may lower DBP below 65 mm Hg, considered potentially harmful for an older patient (65). Similarly, our results are in accord with guidelines from the European Society of Cardiology, which recommended a higher threshold for treatment (SBP ≥150/90 mmHg) in patients older than 80 years (66).

As suggested by others (67, 68), vulnerability to the effects of hypertension may differ by cognitive domain, which would argue for administering tests assessing different cognitive function domains as reported here. In our study, sex differences were observed in specific test performance, such that blood pressure and hypertension were associated with poor performance on category fluency in men and Trails B in women. Given the previously reported sex differences in cognitive function previously reported (38-44), it is plausible that men and women differ in their cognitive vulnerability to high blood pressure and hypertension. Some of the divergent findings in previous reports may be explained by differences in age or cognitive function tests, or lack of sex-specific analyses.

In this study, almost half the men and women (41% and 44%, respectively) were using antihypertensive medication. While some report that antihypertensive medication use is protective against cognitive impairment (69-72), others report higher risk of incident cognitive impairment with antihypertensive drug use, especially diuretics, angiotensin converting enzyme (ACE) inhibitors, and beta blockers (73, 74) as well as a slower decline in cognitive function among patients with dementia (75). Among 302 older adults treated for hypertension, Spinelli et al. (76) found those with poor blood pressure control had lower MMSE and verbal fluency scores. In the present study, there were few differences, however, between those with controlled and uncontrolled hypertension compared to normotensive individuals. Therefore, it is unlikely that associations of blood pressure and hypertension with poor performance observed here are due to the risks or benefits of antihypertensive medication per se.

Several limitations and strengths of this study were considered. Rancho Bernardo Study participants are predominantly Caucasian, well-educated, and have access to medical care, as reflected by the wide use of antihypertensive medications, which may limit generalizability. Because this study is cross-sectional, causality cannot be assumed, although it seems unlikely that the stress or irritation of poor cognitive function raises blood pressure. Furthermore, blood pressure was measured prior to cognitive function assessment during the clinic visit. Given the number of statistical tests performed, the possibility that observed differences were due to chance cannot be excluded. However, all comparisons were a priori attempts to understand the contradictory literature on blood pressure and cognitive function, and the sex-specific associations with blood pressure were consistently observed for the same tests. We also cannot exclude participation bias, as in any sample of older adults where those with the most impaired cognitive function do not participate. However, this bias would yield conservative estimates of any true association, which might explain the absent association in those aged 80 and older. High blood pressure increases the risk of mortality and stroke, further diminishing the influence of those at highest risk of cognitive decline. Finally, no brain imaging studies were performed at this visit, so we are unable to relate changes in cognitive function associated with hypertension to underlying changes in the brain.

This study has several strengths however, including the very high participation rate for community-dwelling seniors, standardized blood pressure measurement, and verified medication use. The homogeneity of this cohort is also an advantage as there is less confounding of test performance due to socio-cultural differences, education, or access to health care.

In conclusion, higher blood pressure and hypertension were associated with poor performance on specific cognitive function tests, which differed by sex. Hypertensive women had poorer performance on Trails B, a test of visuomotor tracking and executive function, whereas hypertensive men showed poorer performance on category fluency, a test of verbal fluency, and semantic memory. These associations were observed in those younger than 80 years of age, but not in older individuals. The absent association of blood pressure with cognitive impairment in those aged 80 and older is compatible with the thesis that older people may require higher blood pressure to maintain cerebral perfusion and could have important clinical implications, or reflect participation bias. Additional studies are needed to prospectively examine the age and sex-specific associations of cognitive function by blood pressure status in older adults.

Abbreviations

ADRC

Alzheimer’s Disease Research Center

BMI

body mass index

DBP

diastolic blood pressure

MMSE

Mini-Mental State Examination

PP

pulse pressure

SBP

systolic blood pressure

Trails B

Trail-making Test, part B

UCSD

University of California, San Diego