The Role of Functional Status on the Relationship between Blood Pressure and Cognitive Decline: The Cardiovascular Health Study.

Introduction

Hypertension or high blood pressure (BP) measured in midlife is associated with negative health outcomes including cognitive decline, cardiovascular disease (CVD), and mortality.^1–4 However, the findings are less clear among some populations of older adults, especially frail/low-functioning persons or the very old.⁵ For instance, randomized control trials have shown inconsistent results regarding benefits from antihypertensive medication on cognitive decline^5,6 and mortality,^7,8 with some studies finding that antihypertensive use reduces the risk of cognitive decline and Alzheimer’s disease, while other studies found no association. Preliminary results from the Systolic Blood Pressure Intervention Trial (SPRINT) MIND trial reported a non-significant reduction in the primary outcome of probable dementia among those participants randomized to intensive blood pressure control,⁹ however results from this study have yet to be published.

We have previously shown that functional status, measured by grip strength and gait speed, and disability in activities of daily living (ADL), modify the relationship between blood pressure and mortality among older adults.^10–13 Among well-functioning older adults, higher blood pressure is associated with worse health outcomes, whereas among those with poor functional status, lower blood pressure appears harmful, potentially due to poor health or decreased perfusion of vital organs. Other studies have shown that higher BP in the oldest old is associated with higher cognitive scores,^14–16 and that this relationship differs by functional status.^14,16 A previous study in the Cardiovascular Health Study (CHS) found that hypertension was not associated with incident dementia, however, this study did not evaluate interactions with functional status. ¹⁷

The objective of this study was to evaluate whether the relationship between systolic and diastolic BP with cognitive decline and incident dementia differed among older adults with and without ADL limitations. We hypothesized that among participants without ADL limitation, elevated BP would be associated with worsening cognitive function over follow-up and a higher risk of incident dementia, whereas among those with ADL limitation the association would be attenuated or in the opposite direction.

Methods

Study Sample

The CHS is a population-based, prospective cohort study designed to examine risk factors for CVD in older adults. The study enrolled 5,201 adults 65 years of age and older in 1989–1990 from four field sites throughout the U.S. An additional sample of 687 African American participants were recruited in 1992–1993.

Participant recruitment has been previously described ¹⁸. Briefly, participants were recruited from Medicare eligibility lists at four field sites: University of California, Davis in Sacramento County, California, Johns Hopkins University in Washington County, Maryland, Wake Forest University School of Medicine in Forsyth County, North Carolina, and University of Pittsburgh in Pittsburg Pennsylvania. Eligibility included all persons living in the household of the individual sampled, who were non-institutionalized, who expected to remain in the area for three years, and who could give informed consent without the need for a proxy. At baseline (1989–1990) and annually during the first 10 years, participants completed clinic visits, physical examinations, health questionnaires, and donation of blood specimens.

Prior studies of the paradoxical relationship between elevated BP and positive health benefits are most commonly observed in adults 75+,^13,14,16 thus the analytical sample was restricted to participants 75 years and older in 1992–1993, yielding a total sample size of 2,097. The baseline year of 1992–1993 was chosen to account for participants with data on dementia.

Blood Pressure

Systolic and diastolic BP were measured in 1992–1993, and were defined as the average of two standard BP readings. Both systolic and diastolic BP were modeled as a continuous variable (per 1 standard deviation (SD) increase in mmHg), and a dichotomous variable representing elevated BP (<140 and ≥ 140 mmHg for systolic BP; <80 and ≥80 mmHg for diastolic BP). Due to the small sample size of participants with baseline DBP of ≥ 90 (n=96), we have used the cut-point of 80 as has been done previously in this cohort.¹³ Although the European Society of Cardiology (ESC) and the European Society of Hypertension (ESH) use a higher threshold of DBP ≥ 90 to define hypertension, ^19,20 the small sample size above this level prohibits us from exploring it further.

Cognitive Function

Cognitive function was measured using two instruments from 1992–1993 for up to 6 years of follow-up. The primary measure of cognitive status was the Modified Mini Mental State Examination (3MSE), which evaluates global cognitive function, change was defined as the slope of the scores over 6 years. Higher scores represent better cognitive functioning. To reduce bias from missing data, the Telephone Interview for Cognitive Status (TICS) was used to convert missing values for 3MSE scores during follow-up years 2–4 described by Arnold et al. ²¹. The secondary measure was executive cognitive ability and processing speed, and were categorized based on the Digit Symbol Substitution Test (DSST), which was also administered annually through 1999 with a score ranging from 0–90, where higher scores represent better cognitive function ²². Cognitive decline using DSST was similarly defined as the slope of scores over 6 years of follow-up.

Incident Dementia

Between 1991–1994, 3660 participants received magnetic resonance imaging (MRI). Of those, 3,602 participants completed the 3MSE at the time of the MRI and were included in the CHS Cognition Study where they were followed until 1999 for incident dementia. Prevalent dementia was defined as being classified with dementia at the time of the MRI. Incident dementia was defined as having been diagnosed with dementia between the fourth follow-up (1994 −1995) and the ninth follow-up visit (1998 – 1999). Protocol for the CHS Cognition Study has been described elsewhere.^23,24 Of the 3,602 participants with MRI data, 1,443 who were 75 years of age or older were included in analyses. There were 164 participants with prevalent dementia at the time of the MRI that were excluded for the incident dementia analyses.

Effect modifiers

Participants were asked whether they had difficulty or were unable to perform any of six ADLs, including walking around the home, getting out of bed, eating, dressing, bathing, and using the toilet. ADL difficulty was dichotomized as none and 1 or more as was previously defined in this cohort ¹³

Potential confounders

Covariates were measured in 1992 – 1993. Demographic variables included clinic site (University of California, Davis, Johns Hopkins University, Wake Forest University, University of Pittsburgh), sex (male, female), race (white/other, black), and education (less than high school, high school or GED, some college or higher degree). Behavioral variables include smoking status (never, former, current), alcohol use defined as none (0 drinks per week), some use (1–7 drinks per week for women, or 1–14 drinks for men), or frequent use (> 7 drinks per week for women, or > 14 drinks per week for men), and body mass index (BMI), which was measured continuously and calculated as weight (kg)/height (m)². Clinical variables included diabetes categorized as normal, impaired fasting glucose (110–125) and diabetic (taking insulin or oral hypoglycemic and insulin level is >=126). LDL-, HDL-, and total cholesterol measured continuously as mg/dL, anti-hypertensive medication use, and apolipoprotein E ɛ4 allele (APOE). Measurement methods for APOE have been published elsewhere.⁹ APOE was dichotomized in this study as having at least one ɛ4 allele (yes, no).^25,26 Kidney disease was measured using serum cystatin C-based estimated glomerular filtration rate (eGFRcys), and was calculated using the Chronic Kidney Disease Epidemiology Collaboration equation. ²⁷

Statistical Analysis

We summarized characteristics at baseline (1992–1993) by ADL limitation (any vs. none). We used t-tests and chi-squared (χ²) tests to evaluate differences between participants with and without ADL limitation. We visually examined the trajectory of both 3MSE and DSST and determined them to be linear. The slopes of 3MSE and DSST were determined through unadjusted linear mixed models, and were used as outcome measures in standard linear regression. We tested effect modification by ADL limitation with systolic and diastolic BP separately by including an interaction term between ADL limitation and BP. Both systolic and diastolic BP were modeled continuously and in categories. Unadjusted and adjusted models were used in ADL difficulty stratified analysis. We used Cox proportional hazards regression to evaluate the relationship between systolic BP and diastolic BP with time to incident dementia stratified by difficulties in ADL.

As a sensitivity analysis, we evaluated the relationship between elevated systolic and diastolic BP with cognitive decline by ADL status and baseline antihypertensive use (yes or no). Additionally, we conducted competing-risk regression with death as the competing-risk using the Fine and Gray method.²⁷

All analyses were performed in Stata (version 15.1, Stata Corporation). Significance was defined for all analyses as p < 0.05.

Results

Participant Characteristics

Among 2,097 adults aged 75 years and older, approximately 16% (n=326) reported limitations in at least one ADL at baseline (1992–1993). Participants without ADL limitations were younger, more likely to be male and white, more likely to have some college or higher degree, have a lower BMI, and have better scores on DSST and 3MSE than those with ADL limitations, while they were more likely to be frequent alcohol users and have diabetes (Table 1). Smoking status, systolic and diastolic BP, hypertension, LDL-HDL- and total cholesterol, and APOE e4 allele status were similar between participants with and without limitations in ADL. Participants without ADL limitations had a mean 3MSE score of 87.24 (SD = 11.96), while the mean baseline 3MSE score for participants with ADL limitation was 83.09 (SD = 14.55). Likewise, participants without ADL limitations had a mean DSST score of 34.22 (SD = 13.10), and those with limitations had a mean baseline DSST score of 28.99 (SD = 13.09).

Table 1.

Characteristics of Participants 75 years of age and older at baseline (1992/1993) in the Cardiovascular Health Study.

Characteristics	No ADL Limitation (n=1,771)	ADL Limitation (n=326)	P-value
	Mean (SD), or N(%)
Age	80.09 (0.09)	81.87 (0.27)	< 0.001
Female	964 (54.43%)	225 (69.02%)	< 0.001
African American (vs. White or others)	244 (13.78%)	61 (18.71%)	0.02
Education			0.01
Less than High School Diploma	531 (29.98%)	131 (40.18%)
High School Diploma	448 (25.30%)	65 (19.94%)
Some College or Higher Degree	792 (44.72%)	130 (39.88%)
Alcohol Use			0.004
Never	1,010 (61.74%)	220 (69.84%)
Light	510 (31.17%)	68 (21.59%)
Frequent	116 (7.09%)	27 (8.57%
Current Smoker			0.10
Never	920 (52.01%)	184 (56.44%)
Former	732 (93.39%)	115 (35.28%)
Current	117 (6.61%)	27 (8.28%)
Body Mass Index (BMI)	25.78 (4.21)	27.69 (3.34)	< 0.001
Systolic BP (mmHg)	139.74 (22.80)	141.11 (23.92)	0.33
Diastolic BP (mmHg)	70.05 (12.18)	68.62 (12.01)	0.06
Hypertension			0.77
Normotensive	663 (38.13%)	113 (35.99%)
Pre-hypertensive	296 (17.02%)	55 (17.52%)
Hypertensive	780 (44.85%)	146 (46.50%)
Total Cholesterol	205.73 (40.09)	208.62 (42.93)	0.29
LDL Cholesterol	125.09 (33.48)	125.63 (37.66)	0.59
HDL Cholesterol	53.32 (14.13)	54.08 (14.72)	0.79
Diabetes			< 0.001
Normal	1,226 (75.26%)	169 (62.83%)
Impaired Glucose	169 (10.37%)	33 (12.27%)
Diabetic	234 (14.36%)	67 (24.91%)
APOE	369 (23.34)	56 (19.51%)	0.16
DSST	34.22 (13.10)	28.99 (13.09)	< 0.001
3MSE	87.24 (11.96)	83.09 (14.55)	< 0.001

Note. ADL = Activities of Daily Living; BP = Blood Pressure; APOE = apolipoprotein E ɛ4 allele; DSST = Digit Symbol Substitution Test; 3MSE = Modified Mini Mental State Exam

Note. Frequent alcohol use was defined as (> 7 drinks per week for women, or > 14 drinks per week for men.

Note. Difficulties in ADL was defined as difficulties with at least one of the following: walking around the home, getting out of bed, eating, dressing, bathing, and using the toilet

Association of BP with Cognitive Decline across Functional Status

Over 6 years of follow-up, the mean annual decline in 3MSE was 1.63 points. The associations between systolic and diastolic BP and cognitive decline varied across ADL limitation status in both unadjusted and adjusted models. In adjusted models, we observed statistically significant effect modification by ADL limitation when evaluating per 1 SD higher in systolic BP (p=0.01), and when using the clinical cut-point for elevated BP (systolic BP ≥ 140 mmHg, p<0.001; diastolic BP ≥ 80 mmHg; p=0.01).

Among participants with no ADL limitations, 1 SD higher in systolic BP was associated with a 0.05-point decrease (9%% CI: 0.09, −0.002) compared with a 0.20-point increase in 3MSE per year over follow-up among participants with ADL limitations (95% confidence interval [CI]: 0.08, 0.32; Table 2). Similarly, among participants without ADL limitations, elevated systolic BP was significantly associated with a 0.15-point decrease in 3MSE scores per year (95% CI: −0.24, −0.07) compared with a 0.30 increase in 3MSE scores per year (95% CI: 0.06, 0.55). Elevated diastolic BP (≥80 mmHg) was associated with an increase in cognitive function in all participants, although the increase was greater in those with ADL limitation than in those without ADL limitations (0.47 points per year vs. 0.18 points per year, p-value for interaction =0.01). The patterns for the outcome of change in DSST were less clear. The association of BP with improvement in DSST over time was stronger among those with ADL limitations than those without, when BP was dichotomized as systolic BP ≥ 140 mmHg and diastolic BP ≥ 80 mmHg (p-values for interaction <0.001 for both). However, the associations were attenuated in adjusted models. When BP was examined as a linear variable, the association between BP and cognitive decline differed by ADL status (p-value for interaction 0.01 for systolic BP and 0.001 for diastolic BP; Supplemental Table 1) with greater improvements in cognitive scores among the ADL limitation group.

Table 2.

The association of blood pressure and annual change in Modified Mini Mental State Exam (3MSE) among Cardiovascular Health Study in participants 75 years of age and older

ADL	No ADL Limitationa (n=1,771)	ADL Limitationb (n=326)	P for
	β Coefficient (95% CI)		Interaction
Systolic BP per
SD mmHg
Unadjusted	−0.11 (−0.14, −0.07)***	0.17 (0.07, 0.26)***	< 0.001
Adjustedc	−0.03 (−0.08, 0.01)	0.16 (0.04, 0.28)**	0.04
Diastolic BP per
SD mmHg
Unadjusted	−0.01 (−0.05, 0.03)	0.14 (0.04, 0.24)**	0.003
Adjustedc	0.06 (0.01, 0.10)*	0.13 (0.004, 0.26)*	0.12
Systolic BP per
≥140 mmHg
Unadjusted	−0.23 (−0.31, 0.16)***	0.33 (0.13, 0.53)**	< 0.001
Adjustedc	−0.15 (−0.24, −0.07)***	0.30 (0.06, 0.55)*	< 0.001
Diastolic BP per
≥80 mmHg
Unadjusted	−0.04 (−0.13, 0.06)	0.45 (0.19, 0.71)**	< 0.001
Adjustedc	0.18 (0.08, 0.29)**	0.47 (0.16, 0.78)***	0.01

Note. ADL = Activities of Daily Living; BP = Blood Pressure

Note. Systolic and Diastolic blood pressure was measured at follow-up visit 3, cognitive decline was measured as the slope of 3MSE between follow-up visit 3 (1992/1993) and follow-up visit 9 (1998/99).

^aNo difficulties in any ADL.

^bDifficulties with at least one of the following: walking around the home, getting out of bed, eating, dressing, bathing, and using the toilet.

^cAdjusted by education level, sex, race, study location, BMI, smoking status, alcohol use, diabetes, antihypertensive medication, LDL cholesterol, HDL cholesterol, total cholesterol, cystatin-C based estimated glomerular filtration rate (eGFRcys) and apolipoprotein E ɛ4 allele.

^*P < 0.05;

^**P < 0.01;

^***P < 0.001

We did not find a significant interaction between antihypertensive medication use, ADL status, and elevated systolic and diastolic BP in either the unadjusted or adjusted models (p=0.10; p=0.99 for systolic BP; p=0.47; p=0.29 for diastolic BP, respectively). In general, we observed similar patterns of effect modification of BP by ADL status among persons on and not on antihypertensive medication. Higher BP was associated with improvements in cognitive function among persons with ADL limitation, whereas this association was attenuated or reversed among those without ADL limitations (Supplemental Table 2).

Association of BP with Incident Dementia across Functional Status

Among 1,279 participants aged 75 years or above with an MRI and data on ADL status, there were 308 cases (49 per 1,000 person-years) of incident dementia over 6 years of follow-up. We did not find significant effect modification of systolic or diastolic BP by ADL status and incident dementia when measured per 1 SD higher in baseline BP in the unadjusted or adjusted models, or with elevated BP (Table 3). We observed a significant interaction with elevated systolic BP in the unadjusted interaction model (p=0.02); when stratified, we observed a hazard ratio (HR): 1.33 (1.04, 1.70) among those without limitations, and an HR: 0.62 (95% CI: 0.34, 1.12) among those with ADL limitations. This effect modification was attenuated after adjustment. In general, among participants with difficulties in ADL, the risk of incident dementia associated with higher blood pressure tended to be lower than among participants reporting no difficulties, providing suggestive evidence of patterning in the same direction compared with the trends observed using cognitive change as the outcome.

Table 3.

The association of elevated blood pressure and time to incident dementia among Cardiovascular Health Study participants 75 years of age and older .

ADL	No ADL Limitationa (n=1,114)	ADL Limitationb (n=141)	P for
	Hazard Ratio (95% CI)		Interaction
Systolic BP per
SD mmHg
Unadjusted	1.13 (1.00, 1.28)	0.87 (0.65, 1.16)	0.10
Adjustedc	1.09 (0.94, 1.26)	0.95 (0.63, 1.43)	0.36
Diastolic BP per
SD mmHg
Unadjusted	0.99 (0.86, 1.12)	0.89 (0.64, 1.22)	0.56
Adjustedc	0.98 (0.83, 1.14)	0.85 (0.51, 1.41)	0.48
Systolic BP per
≥140 mmHg
Unadjusted	1.33 (1.04, 1.70)*	0.62 (0.34, 1.12)	0.02
Adjustedc	1.21 (0.90, 1.63)	0.80 (0.34, 1.89)	0.14
Diastolic BP per
≥80 mmHg
Unadjusted	1.10 (0.81, 1.51)	0.67 (0.27, 1.71)	0.33
Adjustedc	0.92 (0.62, 1.36)	0.97 (0.27, 3.45)	0.81

Note. ADL = Activities of Daily Living; BP = Blood Pressure; CI = Confidence Interval

Note. Systolic and Diastolic blood pressure was measured at follow-up visit 3, incident cognitive impairment was measured between follow-up visit 4 (1994/1995) and visit 9 (1999)

Note. There were 308 cases (49 per 1,000 person-years) of incident dementia over 6 years of follow-up.

^aNo difficulties in any ADL.

^bDifficulties with at least one of the following: walking around the home, getting out of bed, eating, dressing, bathing, and using the toilet

^cAdjusted by education level, sex, race, study location, BMI, smoking status, alcohol use, diabetes, antihypertensive medication, LDL cholesterol, HDL cholesterol, total cholesterol, cystatin-C based glomerular filtration rate (eGFRcys) and apolipoprotein E ɛ4 allele.

^*P < 0.05;

^**P < 0.01;

^***P < 0.001

We observed similar patterns of the relationship between blood pressure and incident dementia when using competing risk regression; no effect modification by ADL limitation was observed (Supplemental Table 3).

Discussion

Among 2,097 community dwelling older adults, we found that the association between BP and cognitive decline varied by ADL limitation. Among older adults without ADL limitations, higher systolic and diastolic BP was associated with a decreases or small increase in 3MSE scores over time. On the contrary, among older adults with ADL limitation, higher systolic and diastolic BP were associated with an increase in 3MSE scores over follow-up. Risk of incident dementia did not differ by ADL status. The relationship with DSST was less clear, suggesting inconsistencies among cognitive measures with elevated BP. While effect sizes were small, our findings suggest that functional status, assessed by difficulties in ADL, may be a useful measure to identify older adults who are at most risk for cognitive decline associated with high BP.

Our findings are consistent with previous research showing that elevated BP in older adults with poor functional status is associated with greater health benefits compared to frail older adults with lower levels of BP. For example, using the National Health and Nutrition Survey (NHANES), we found that in participants that did not complete the walk test, elevated BP was strongly associated with a lower risk of death compared to slow and fast walkers.¹¹ Using the Health and Retirement Study, we found that elevated systolic and diastolic BP were associated with a 6% and 16% lower mortality rate among those with weak grip strength compared to those with normal grip strength.¹⁰ Finally, in the CHS, results showed that among participants with difficulties in ADL, elevated diastolic BP was associated with reduced risk of cardiovascular disease and a lower risk of death.¹³

We extended previous research by examining the longitudinal relationship of BP with cognitive decline and incident dementia. Similar to our study, Ogliari et al. found a modifying effect of functional status in the cross-sectional relationship between BP and cognitive function, showing that higher BP was associated with higher cognitive scores among participants with dependence in ADLs.¹⁴ Moreover, Sabayan et al. found that among older adults with difficulties in ADL, higher levels of SBP was associated with lower annual declines in the mini mental state exam (MMSE) compared to participants not reporting difficulties in ADL.¹⁶ While we did not find statistical evidence that the relationship between BP and incident dementia differed by functional status, the trend that we observed was comparable to a study performed by Verghese et al., which found that low diastolic BP was associated with an increased risk of incident dementia among participants 75 and older.²⁸

While the mechanism between high BP and improved cognitive function it is not entirely clear, one explanation includes the belief that higher BP is important for maintaining adequate perfusion of vital organs, such as the brain.^16,29 For instance, hypotension and treatment for hypertension may increase hypoperfusion in the brain, which can result in hypoxia and increase the risk of neurodegenerative disorders.³⁰ In our study, persons with ADL limitation had worse scores on the cognitive function tests, and were likely to be further along in the cognitive disease process. These people may have diminished cognitive reserve and be at elevated risk for hypoperfusion and other vascular stressors.

Strengths of this study include the large, heterogeneous cohort of community-dwelling older adults, and the longitudinal design, expanding on results observed in cross-sectional studies.³¹ There are some limitations in our study that should be considered. One limitation includes the small sample size for participants with MRI data, which may have limited our ability to detect a statistically significant relationship between BP and incident dementia. Another limitation is the inherent observational design, limiting our ability to make causal inferences about the nature of BP and cognitive decline by functional status. Another limitation is the possibility that changes in cognitive function are a result of regression to the mean or loss to follow-up. However, we attempted to address this concern by completing competing risk regression, which did not substantially alter our findings.

Our findings are noteworthy because the recent 2017 American College of Cardiology and American Heart Association (ACC/AHA) Hypertension Guidelines recommend a BP target of < 130/80 mmHg for all adults 65 and older. Further research will help to determine whether our findings contrary to this recommendation are observed in randomized controlled trials of diverse elders. Trials have historically underrepresented frail older adults, older adults ≥ 75 years of age, or older adults with comorbidities. This makes it challenging to disentangle the contributions of unmeasured confounding in observational studies and selection bias in trials in explaining these apparently conflicting results.

Conclusion

In conclusion, we found modest, but statistically significant differences in the relationship between BP and cognitive decline by functional status in a large cohort study of US adults aged 75 years and older. Elevated BP (>140/80 mmHg) was associated an expected worsening or no change in cognitive function among functioning older adults, and was associated with improvement in cognitive function among older adults with ADL limitations. Additional studies should be examined to evaluate the clinical importance of the observed changes in cognitive status and its association with BP. Additional research is needed to evaluate if older adults with functional disabilities may find cognitive benefit from maintaining elevated BP levels.