Blood Pressure and Cognitive Decline in Prevalent Hemodialysis Patients

Abstract

Background

Hypertension is associated with cognitive decline in the general population. It is unclear what impact blood pressure has on cognitive decline in patients receiving maintenance hemodialysis.

Methods

Using a longitudinal cohort of 314 prevalent hemodialysis patients without dementia at baseline, we examined the association of pre-dialysis systolic and diastolic blood pressure, pulse pressure, and intra-dialytic systolic blood pressure change (pre minus post), averaged for a month, with cognitive decline. Cognitive function was determined by a neurocognitive battery, administered yearly. Individual cognitive test results were reduced into two domain scores using principal components analysis (by definition mean of zero and standard deviation of one), representing memory and executive function. Joint models, allowing for characterization of cognitive score slopes and including adjustment for potential confounders, were utilized to account for competing risks from death, dropout, or kidney transplantation.

Results

Mean age was 62 years; 54% were men, 23% were black and 90% had at least a high school education. During median follow up of 2.1 years (25^th-75^th: 1.0–4.5), 191 had at least one follow up test, 148 died and 43 received kidney transplants. Low pre-dialysis diastolic blood pressure and high pulse pressure were both associated with steeper executive function decline (each 10 mmHg lower diastolic blood pressure = −0.03 SD (−0.01, −0.05) per year steeper decline in executive function; each 10 mmHg higher pulse pressure = −0.03 SD (−0.06, −0.01) steeper decline) but not for memory function. Systolic blood pressure and intra-dialytic change were not associated with steeper decline for either memory or executive function.

Conclusions

No relationship was seen between systolic blood pressure or intra-dialytic change in blood pressure with cognitive decline. In prevalent hemodialysis patients, lower pre-dialysis diastolic blood pressure and wider pre-dialysis pulse pressure are associated with steeper cognitive decline in executive function but not memory.

Introduction

Studies of blood pressure in the general population demonstrate associations between higher systolic and diastolic blood pressure and cognitive impairment.^1–4 Patients with chronic kidney disease (CKD) often have concomitant hypertension, potentially placing them at high risk for cognitive impairment.⁵ Those with kidney failure requiring dialysis have an even higher prevalence of cognitive impairment with 20–60% demonstrating cognitive deficits,⁶ with the majority of dialysis patients also displaying hypertension.

There is significant controversy over appropriate blood pressure targets in the general population.⁷ This uncertainty extends to patients with kidney disease⁸, a population that may benefit from intensive blood pressure targets⁹, while also being more susceptible to adverse outcomes from aggressive blood pressure control.¹⁰ Additional ambiguity arises from the aging related fall in diastolic blood pressure and associated loss of vascular elasticity, which raises the question as to the clinical impact of low diastolic blood pressure. Unfortunately, minimal data are available to support blood pressure targets in hemodialysis (HD) populations.¹¹ Importantly, there are only a few studies of blood pressure and cognitive impairment in patients with CKD^5,12, and none have evaluated the longitudinal relationship between specific blood pressure measures and changes in cognitive function.

We previously demonstrated an association between low diastolic blood pressure and high pulse pressure with impaired executive function (a cognitive domain closely linked with cerebrovascular disease) in cross sectional analyses; however these associations were attenuated after multivariable adjustment.¹² Based on these cross-sectional findings and previously described longitudinal associations between blood pressure and cognitive function in non-dialysis populations, we evaluated the association between pre-dialysis systolic and diastolic blood pressure as well as pulse pressure with cognitive decline in a cohort of maintenance hemodialysis patients, focusing on memory and executive function cognitive domains.^6,13

Materials and Methods

Study Population

Outpatients eighteen years old or older receiving maintenance in-center hemodialysis at five Dialysis Clinic Inc. (DCI) units and one hospital-based outpatient unit (St. Elizabeth’s Medical Center) in the greater Boston area were screened for the Cognition and Dialysis Study, with study enrollment occurring from January 28, 2004 to May 31, 2012.⁶ Eligibility criteria included English fluency as well as sufficient visual and hearing acuity to complete neurocognitive testing. To minimize floor effects and to maximize the likelihood of completing the full battery of tests, individuals with MMSE score ≤10 and/or possible diagnosis of dementia were excluded. Determination of whether a participant had possible dementia was based on a detailed review of the dialysis unit and hospital electronic record for any mention of dementia. Additional exclusion criteria included non-access related hospitalization within one month of screening, receipt of hemodialysis for less than one month, and single pool (sp) Kt/V <1.0. The Tufts Medical Center/Tufts University Institutional Review Board approved this study, and all participants who agreed to undergo detailed cognitive testing signed informed consent. The clinical and research activities reported are consistent with the Declaration of Helsinki.

Baseline Demographics and Clinical Characteristics

Demographic, clinical and laboratory factors were ascertained at the time of cognitive testing. Education (<12^th grade, high school graduate to less than 2 years of college, and ≥ 2 years of college) was obtained via patient questionnaire. History of cardiovascular disease (CVD), defined as a composite of either coronary artery disease and/or peripheral vascular disease, was determined by patient history or documentation in the patient’s electronic or paper chart. Patients were queried about personal history of myocardial infarction and coronary revascularization, which were used to define coronary disease, and about intermittent claudication and peripheral vascular disease, which were used to define peripheral vascular disease. Additionally, DCI electronic medical and paper records were reviewed for a history of these conditions, with specific focus on problem lists, hospital discharge summaries, cardiac test results, and procedure results. Additional medical history including primary cause of ESRD, hemodialysis vascular access type, and dialysis vintage (time since hemodialysis initiation) were obtained from the DCI or St. Elizabeth’s electronic record. Serum albumin and spKt/V most proximate to the time of cognitive testing were obtained from participant medical records.

Blood Pressure

For all blood pressure measures, the monthly average at the time of baseline cognitive testing was utilized. Systolic and diastolic blood pressure readings were obtained in the outpatient dialysis unit as a part of usual clinical care immediately prior to the start as well as immediately after completion of hemodialysis sessions were obtained from the DCI electronic records. Per dialysis unit protocols, nurses and technicians are instructed to measure blood pressure in the non-access arm with an appropriately sized cuff, using an automated system, prior to the initiation of the hemodialysis session. Pulse pressure was obtained by subtracting diastolic blood pressure from systolic blood pressure. Intra-dialytic change in blood pressure was defined as the difference between pre-dialysis systolic blood pressure and post-dialysis systolic blood pressure.

Neurocognitive Assessment

At study enrollment, participants were administered a battery of neurocognitive tests by research coordinators after a period of training and direct observation by the study neuropsychologist (Dr. Scott). The same battery of tests was administered yearly to study participants whenever possible. To maintain quality and inter-rater reliability, testing was observed by the study neuropsychologist at 3–6 month intervals. To limit subject fatigue, all testing was completed during the 1^st hour of hemodialysis. Using the same battery of tests, we have previously demonstrated similar performance regardless of whether testing was performed during the first hour of dialysis or before the start of dialysis.¹⁴ When possible, neurocognitive testing was performed in a private room or in as quiet an environment as possible. The neurocognitive battery included well-validated commonly used cognitive tests (Supplemental Table 1) that possess high inter- and intra-rater reliability. The MMSE was used as a screening test.¹⁵ The neurocognitive battery consisted of the Wechsler Memory Scale-III (WMS-III) Word List Learning Subtest,¹⁶ the Wechsler Adult Intelligence Scale-III (WAIS-III) Block Design¹⁶ and Digit Symbol-Coding Subtests,¹⁶ and Trail Making Tests A and B (Trails A and B).¹⁷ For Trails B, a 300 second time limit was imposed, with those unable to complete the test during this time period considered “non-completers”. In year 3 of the study, the cognitive panel was expanded to include additional verbal tests assessing both memory and executive functions, including Digit Span (forwards and backwards),¹⁶ the Mental Alternation Test,¹⁸ and the Controlled Oral Word Association Test (COWAT).¹⁹

Our pre-specified primary outcomes were the association of each baseline blood pressure measure with change in memory and executive function over time. Principal component analysis (PCA) with varimax rotation was used as a data reduction technique with 292 participants to derive composite scores for separate cognitive domains (memory or executive function) in the entire study population.²⁰ For 18 individuals who were missing baseline results on one cognitive test (or 2 results if derived from the same test), single item imputation was performed using multivariable linear regression models based on performance on other tests in the cognitive battery. Two principal components with eigenvalues greater than two were obtained (Component 1 eigenvalue = 2.87, Component 2 eigenvalue = 2.29), and the resulting component scores subsequently were used as co-primary outcomes. Using this method, all component scores have a mean of zero and standard deviation of one. The first component was interpreted to reflect executive functioning, attention, and processing speed (referred to as executive function in the Results section), with the Trails A and B, Block Design, and Digit Symbol-Coding tests contributing significantly (Supplementary Table 1). The second component primarily was composed of Word List Learning Recall and Recognition and was interpreted to reflect memory. Component loadings for deriving the principal component score at the baseline examination (n = 292) were used to calculate the principal component scores for follow-up testing, which included 8 patients who did not have enough data to calculate PCA scores at baseline, bringing the number of unique participants up to 300. Digit Span, Mental Alternations and the COWAT were not used to calculate the PCA because of the smaller number of individuals who completed these tests. In exploratory analyses, we also examined he association of each blood pressure measure with individual test scores; the goal of these analyses was to better understand which component tests contributed the most to any observed associations.

Statistical analysis

Descriptive characteristics of the study population were described by quartiles of diastolic blood pressure with frequency count and proportions for categorical and binary variables, means with standard deviations for continuous normally distributed variables, and medians with interquartile ranges for skewed variables.

We used random effects mixed models, accounting for the correlation between repeated measures, to explore the change in cognitive test scores over time. We postulated a linear model with a random intercept and slope for each cognitive test in which at time j for the i^th patient, the cognitive test Y_ij is given by, ${\mathrm{Y}}_{\mathrm{i}\mathrm{j}}=({\beta }_1+{\mathrm{b}}_{1\mathrm{i}})+({\beta }_2+{\mathrm{b}}_{2\mathrm{i}}){\mathrm{t}}_{\mathrm{j}}+{\epsilon }_{\mathrm{j}\mathrm{j}};i=1;\dots ;\mathrm{n};\mathrm{j}=1;\dots ;{\mathrm{n}}_i$ where β₁ and β₂ are the population mean intercept and rate of change. Because of the high rate of dropout due to mortality and transplant, and the fact that mortality and other causes of dropout may be informative, such that cognitive decline may have been greater in those who die before subsequent testing, we used shared random effects joint models for repeated measures and time to event data to estimate the rate of change in cognitive function decline adjusting for informative censoring due to mortality, transplant or other reasons.²¹ The joint models approach includes a linear mixed sub-model for the repeated measurement and a survival sub-model for time to dropout. These two models are linked using a shared random effect to account for non-informative dropout. The survival sub-model included was a stratified cause-specific Cox competing risk model with three possible endpoints: (1) mortality, which also included dropout due to prolonged hospitalization or being too sick to complete testing; (2) transplant, transfer to peritoneal dialysis or transfer to a non-participating facility; and (3) no dropout. We separated out these causes of dropout to account for the fact that the association of the most proximate cognitive test with mortality and transplant could be in opposite directions.

To explore the association of each blood pressure measure with the rate of change of the cognitive test, the linear mixed sub-model included the main effect of the covariate of interest, time since baseline as a linear term, and their product. The time effect provides the slope of the cognitive test for the reference level of the covariate, and the interaction term provides the increase/decrease in this slope for the other level of the covariate. Models were initially unadjusted and subsequently adjusted for baseline age, sex, race, education level, cause of ESRD, dialysis vintage and history of CVD. The competing risk survival sub-model was adjusted for the same covariates included in the linear mixed sub-model in addition to variables previously shown to be associated with mortality including vascular access type, serum albumin, BMI and spKt/V.²² To explore potential non-linear relationships, we divided the cohort into quartiles for each blood pressure measure, and examined the relationship between each quartile with cognitive decline. Plots were created to demonstrate the rate of decline for each quartile of blood pressure measures, adjusting for the same covariates described above. Figures represent predicted trajectories for four individuals with levels equaling the mean blood pressure measure within each quartile. Finally, to explore if the effect of blood pressure parameters on the intercept and rate of change of the cognitive test is deviant from linear in favor of a U-shaped effect, we also included in our multivariable model the square of the BP term as an additional main effect and also as an additional interaction term with follow-up time.

Significance was assessed using a two sided alpha of 0.05 and results were presented with 95% confidence intervals. All analyses were performed using SAS Enterprise Guide (Version 7.12, Cary, NC) and R language (version 3.3.1, R Foundation for Statistical Computing, Vienna Austria). Random effects models were fitted using the lme function in R package nlme. The joint models were estimated using the JM package in R (version JM 1.4–4).²³

Results

Baseline Characteristics

Among the 929 patients screened, 414 were ineligible for complete cognitive testing. Reasons included language barriers (n = 194, 47%); behavioral and psychiatric issues (n = 14, 3%); physical impediments such as blindness, hearing loss, and paralysis (n = 62, 15%); recent acute hospitalization or impending death (n = 108, 26%); mention of dementia in the medical record (n = 35, 8%); and a score on the MMSE of less than 10 (n =1, <1%). Of the remaining 515, 314 individuals consented to and underwent detailed cognitive testing.⁶ Three hundred participants had sufficient data to calculate memory and executive principal components. Table 1 describes the demographics and clinical characteristics of the whole study group and by quartiles of diastolic blood pressure. The mean (SD) age of study participants at enrollment was 62 (17) years; 54% were men, 23% were black and 90% had at least a high school education (Table 1). During median follow up of 2.1 years (25^th – 75^th: 1.0 – 4.5), 191 (64%) had at least one follow up test with median number of follow-up tests of 2 (25^th – 75^th: 1 – 4), while 148 (49%) died during the study, 43 (14%) received a kidney transplant and 39 (13%) dropped out because of modality change or transfer to another dialysis unit. Across quartiles of diastolic blood pressure, those with lower diastolic blood pressure were more likely to be older and non-black and were more likely to have history of CVD (peripheral vascular disease and/or coronary disease) and diabetes (Table 1). Other demographics and clinical characteristics were similar across diastolic blood pressure quartiles.

Table 1.

Baseline characteristics by Quartile of Diastolic Blood Pressure

	Total Cohort	Quartile 1	Quartile 2	Quartile 3	Quartile 4	Trend p
	N = 300	N = 74	N = 70	N = 80	N = 76
Pre DBP monthly average (mmHg)	73.5 ± 12.2	58.6 ± 4.5	68.7 ± 2.4	76.6 ± 2.7	89.1 ± 7.7	<.0001
Age (years)	62.4 ± 16.6	71.4 ± 12.0	66.9 ± 14.5	60.2 ± 17.1	51.9 ± 15.3	<.0001
Female	138 (46.0%)	33 (44.6%)	30 (42.9%)	37 (46.3%)	38 (50.0%)	0.45
Black	68 (22.7%)	7 (9.5%)	11 (15.7%)	22 (27.5%)	28 (36.8%)	<.0001
Education						0.81
<12th grade	29 (9.7%)	5 (6.8%)	11 (15.7%)	8 (10.0%)	5 (6.6%)
High school graduate	165 (55.0%)	42 (56.8%)	36 (51.4%)	48 (60.0%)	39 (51.3%)
>2 Years college	106 (35.3%)	27 (36.5%)	23 (32.9%)	24 (30.0%)	32 (42.1%)
Pre SBP monthly average (mmHg)	141.3 ± 20.8	121.9 ± 16.7	137.7 ± 13.7	145.5 ± 16.2	159.0 ± 17.1	<.001
Pulse Pressure (mmHg)	67.8 ± 14.7	63.4 ± 14.5	69.0 ± 13.9	69.0 ± 15.3	69.9 ± 14.3	0.01
Post-Pre SBP (mmHg)	−10.0 ± 23.5	0.2 ± 18.7	−7.9 ± 24.3	−12.3 ± 22.0	−19.5 ± 24.6	<.0001
Stroke	51 (17.0%)	14 (18.9%)	14 (20.0%)	11 (13.8%)	12 (15.8%)	0.42
Peripheral Vascular Disease	69 (23.0%)	26 (35.1%)	19 (27.1%)	12 (15.0%)	12 (15.8%)	0.001
Diabetes	143 (47.7%)	38 (51.4%)	42 (60.0%)	39 (48.8%)	24 (31.6%)	0.007
Heart Failure	106 (35.3%)	29 (39.2%)	30 (42.9%)	25 (31.3%)	22 (29.0%)	0.09
Coronary Artery Disease	111 (37.0%)	43 (58.1%)	32 (45.7%)	21 (26.3%)	15 (19.7%)	<0.0001
Primary cause of ESRD						0.004
Diabetes	102 (34.0%)	33 (44.6%)	29 (41.4%)	24 (30.0%)	16 (21.1%)
Hypertension	57 (19.0%)	13 (17.6%)	15 (21.4%)	12 (15.0%)	17 (22.4%)
Other	50 (16.7%)	7 (9.5%)	7 (10.0%)	18 (22.5%)	18 (23.7%)
Unknown	39 (13.0%)	9 (12.2%)	11 (15.7%)	9 (11.3%)	10 (13.2%)
Glomerulonephritis	52 (17.3%)	12 (16.2%)	8 (11.4%)	17 (21.3%)	15 (19.7%)
Dialysis Vintage (months)	14.4 (7.0, 35.4)	12.7 (6.5, 30.4)	18.1 (8.1, 40.0)	13.4 (6.4, 26.0)	16.9 (6.9, 51.5)	0.25
Vascular Access						0.6
Fistula	194 (64.7%)	47 (63.5%)	49 (70.0%)	55 (68.8%)	43 (56.6%)
Graft	18 (6.0%)	4 (5.4%)	6 (8.6%)	3 (3.8%)	5 (6.6%)
Catheter	88 (29.3%)	23 (31.1%)	15 (21.4%)	22 (27.5%)	28 (36.8%)
Body Mass Index (kg/m2)	28.1 ± 6.5	27.8 ± 6.1	29.7 ± 7.3	28.2 ± 6.2	26.9 ± 6.1	0.22
Kt/V	1.51 ± 0.24	1.53 ± 0.23	1.52 ± 0.23	1.47 ± 0.23	1.53 ± 0.27	0.37
Albumin (g/dL)	3.8 ± 0.4	3.7 ± 0.3	3.8 ± 0.3	3.8 ± 0.4	3.9 ± 0.4	0.002

Data are presented either as n (%), mean ± standard deviation or median (25^th, 75^th percentiles)

Similar results were seen for linear mixed models and joint models. Therefore, as the joint models account for potential informative censoring from death or kidney transplantation, only joint model findings are described below.

Diastolic Blood Pressure

Lower diastolic blood pressure was a significant risk factor for steeper decline in the executive function summary score but not for the memory score (Table 2). Lower pre-dialysis diastolic blood pressure was also a significant risk factor for steeper cognitive decline across multiple cognitive tests (Table 2), the majority of which assess executive function. Similar patterns were seen in quartile analysis (Figure), with the lowest quartile of diastolic blood pressure measures associated with the steepest decline in executive function, while no patterns were observed for memory (Supplementary Figure). For example, after five years a participant in the lowest diastolic blood pressure quartile (average diastolic BP of 58 mm/Hg) would on average have a 0.4 SD lower executive function score compared to a participant in the highest diastolic BP quartile (average diastolic BP of 89 mm/Hg).

Table 2.

Association between diastolic blood pressure (per 10 mmHg higher) and difference in slope of cognitive tests (per year change)

Type of Test	Cognitive test	Scoring	n	N	Unadjusted Slope^	p	Adjusted Slope^ (95% CI)	p
Summary Score	Memory	SD*	300	645	0.01 (−0.01, 0.04)	0.2	0.01 (−0.01, 0.04)	0.2
Summary Score	Executive Function	SD*	300	645	0.03 (0.00, 0.05)	0.03	0.03 (0.00, 0.05)	0.03
Screening	MMSE	No. correct	314	785	0.06 (−0.06, 0.17)	0.3	0.05 (−0.06, 0.16)	0.4
Primarily Memory	Short Delayed Recall	No. correct	312	779	0.01 (−0.06, 0.08)	0.8	0.01 (−0.06, 0.08)	0.8
	Delayed Recall	No. correct	312	774	0.10 (0.04, 0.17)	<0.01	0.10 (0.04, 0.16)	<0.001
	Word Recognition	No. correct	312	775	0.05 (−0.01, 0.11)	0.1	0.05 (−0.01, 0.11)	0.1
Primarily Executive Function	Block Design	No. completed	311	752	0.32 (0.08, 0.56)	<0.01	0.30 (0.07, 0.52)	<0.01
	Digit Symbol Substitution	No. completed	296	676	0.63 (0.29, 0.96)	<0.001	0.58 (0.27, 0.88)	<0.001
	Digits Forward	No. correct	187	452	−0.03 (−0.09, 0.03)	0.3	−0.02 (−0.08, 0.04)	0.4
	Digits Backward	No. correct	186	451	0.03 (−0.03, 0.10)	0.3	0.04 (−0.02, 0.11)	0.2
	Trail Making Part A	Time to completion (sec)	302	693	−1.53 (−3.86, 0.79)	0.2	−1.43 (−3.71, 0.85)	0.2
	Trail Making Part B	Time to completion (sec)	299	681	−2.06 (−4.26, 0.14)	0.07	−1.94 (−3.96, 0.09)	0.1
	Mental Alternations	No. correct	187	453	0.03 (−0.17, 0.22)	0.8	0.08 (−0.11, 0.27)	0.4
	COWAT (Animals)	No. correct	187	453	0.11 (−0.05, 0.28)	0.2	0.14 (−0.02, 0.31)	0.1
	COWAT (Supermarket)	No. correct	187	454	0.22 (0.03, 0.41)	0.02	0.25 (0.07, 0.44)	<0.01

^{^}Slope is the rate of change in the cognitive test per year.

^*Summary scores represent a per standard deviation change, with a mean of zero and SD of 1. Negative coefficients are associated with steeper decline in scores over time except on the Trail Making Tests.

Bold indicates statistically significant differences over time.

n = number of unique patients and N = number of follow up visits. The Digits, Mental Alternations, and COWAT tests were added halfway through study enrollment.

MMSE, Mini-Mental State Examination; COWAT, controlled oral word association test.

Figure 1:

Figures show the predicted trajectories of executive function over time for four modeled subjects with levels equaling the mean blood pressure measure within each quartile. Adjusted for baseline age, sex, race, education level, cause of ESRD, dialysis vintage and history of CVD. The competing risk survival sub-model was adjusted for the same covariates included in the linear mixed sub-model in addition to variables previously shown to be associated with mortality including vascular access type, serum albumin, BMI and spKt/V. Slope values are presented as per 10mmHg change in blood pressure measure.

Systolic Blood Pressure

Higher pre-dialysis systolic blood pressure was not a risk factor for cognitive decline in unadjusted analyses (Table 3). This finding was consistent in multivariable analyses and in quartile analyses (Figures and Supplemental Figure). Systolic blood pressure was similarly not associated with steeper decline in any of the component cognitive tests (Table 2).

Table 3.

Association between systolic blood pressure (per 10 mmHg higher) and difference in slope of cognitive tests (per year change)

Type of Test	Cognitive test	Scoring	n	N	Unadjusted Slope^	p	Adjusted Slope^ (95% CI)	p
Summary Score	Memory	SD*	300	645	0.00 (−0.01, 0.02)	0.5	0.00 (−0.01, 0.02)	0.8
Summary Score	Executive Function	SD*	300	645	0.00 (−0.02, 0.02)	0.9	0.00 (−0.02, 0.01)	0.8
Screening	MMSE	No. correct	314	785	−0.03 (−0.10, 0.04)	0.4	−0.03 (−0.11, 0.04)	0.3
Primarily Memory	Short Delayed Recall	No. correct	312	779	−0.01 (−0.06, 0.04)	0.8	−0.01 (−0.06, 0.04)	0.6
	Delayed Recall	No. correct	312	774	0.03 (−0.01, 0.07)	0.2	0.02 (−0.02, 0.06)	0.2
	Word Recognition	No. correct	312	775	0.01 (−0.04, 0.06)	0.7	0.01 (−0.04, 0.06)	0.8
Primarily Executive Function	Block Design	No. completed	311	752	0.10 (−0.06, 0.26)	0.2	0.08 (−0.07, 0.23)	0.3
	Digit Symbol Substitution	No. completed	296	676	0.11 (−0.14, 0.35)	0.4	0.07 (−0.16, 0.29)	0.6
	Digits Forward	No. correct	187	452	0.01 (−0.04, 0.05)	0.8	0.01 (−0.04, 0.05)	0.8
	Digits Backward	No. correct	186	451	0.02 (−0.03, 0.06)	0.5	0.02 (−0.03, 0.06)	0.4
	Trail Making Part A	Time to completion (sec)	302	693	0.90 (−0.59, 2.40)	0.2	0.97 (−0.50, 2.44)	0.2
	Trail Making Part B	Time to completion (sec)	299	681	0.30 (−1.21, 1.80)	0.7	0.34 (−0.99, 1.68)	0.6
	Mental Alternations	No. correct	187	453	−0.01 (−0.15, 0.13)	0.9	0.02 (−0.12, 0.15)	0.8
	COWAT (Animals)	No. correct	187	453	0.13 (0.02, 0.24)	0.03	0.13 (0.02, 0.24)	0.02
	COWAT (Supermarket)	No. correct	187	454	0.07 (−0.09, 0.22)	0.4	0.07 (−0.07, 0.22)	0.3

^{^}Slope is the rate of change in the cognitive test per year.

^*Summary scores represent a per standard deviation change.

Negative coefficients are associated with steeper decline in scores over time except on the Trail Making Tests.

Bold indicates statistically significant differences over time.

n = number of unique patients and N = number of follow up visits. The Digits, Mental Alternations, and COWAT tests were added halfway through study enrollment.

MMSE, Mini-Mental State Examination; COWAT, controlled oral word association test.

Pulse Pressure

Higher pulse pressure was associated with steeper decline in executive function in both unadjusted and adjusted analyses; this relationship was not observed for memory (Table 4). High pulse pressure was also associated with steeper decline in several individual cognitive tests, including both trailmaking tests. A similar pattern of higher pulse pressure associated with steeper decline in executive function but not memory was seen in quartile analyses (Figure and Supplemental Figure).

Table 4.

Association between pulse pressure (per 10 mmHg higher) and difference in slope of cognitive tests (per year change)

Type of Test	Cognitive test	Scoring	n	N	Unadjusted Slope^	p	Adjusted Slope^ (95% CI)	p
Summary Score	Memory	SD*	300	645	0.00 (−0.03, 0.02)	0.8	0.00 (−0.03, 0.02)	0.8
Summary Score	Executive Function	SD*	300	645	−0.04 (−0.06, −0.01)	<0.01	−0.03 (−0.06, −0.01)	<0.01
Screening	MMSE	No. correct	314	785	−0.13 (−0.24, −0.02)	0.02	−0.13 (−0.24, −0.02)	0.02
Primarily Memory	Short Delayed Recall	No. correct	312	779	−0.03 (−0.11, 0.04)	0.4	−0.03 (−0.11, 0.04)	0.4
	Delayed Recall	No. correct	312	774	−0.05 (−0.13, 0.02)	0.2	−0.05 (−0.12, 0.02)	0.2
	Word Recognition	No. correct	312	775	−0.03 (−0.11, 0.04)	0.4	−0.03 (−0.11, 0.04)	0.4
Primarily Executive Function	Block Design	No. completed	311	752	−0.13 (−0.38, 0.12)	0.3	−0.12 (−0.35, 0.11)	0.3
	Digit Symbol Substitution	No. completed	296	676	−0.50 (−0.88, −0.12)	0.01	−0.51 (−0.87, −0.16)	<0.01
	Digits Forward	No. correct	187	452	0.04 (−0.03, 0.11)	0.3	0.04 (−0.03, 0.10)	0.3
	Digits Backward	No. correct	186	451	−0.01 (−0.08, 0.07)	0.9	0.00 (−0.07, 0.07)	>0.99
	Trail Making Part A	Time to completion (sec)	302	693	3.42 (1.25, 5.58)	<0.01	3.49 (1.33, 5.65)	<0.01
	Trail Making Part B	Time to completion (sec)	299	681	3.24 (0.98, 5.50)	<0.01	2.83 (0.74, 4.93)	<0.01
	Mental Alternations	No. correct	187	453	−0.09 (−0.31, 0.12)	0.4	−0.05 (−0.26, 0.15)	0.6
	COWAT (Animals)	No. correct	187	453	0.15 (−0.03, 0.32)	0.1	0.14 (−0.03, 0.31)	0.1
	COWAT (Supermarket)	No. correct	187	454	−0.14 (−0.38, 0.10)	0.2	−0.11 (−0.31, 0.09)	0.3

^{^}Slope is the rate of change in the cognitive test per year.

^*Summary scores represent a per standard deviation change, with a mean of zero and SD of 1.

Negative coefficients are associated with steeper decline in scores over time except on the Trail Making Tests.

Bold indicates statistically significant differences over time.

n = number of unique patients and N = number of follow up visits. The Digits, Mental Alternations, and COWAT tests were added halfway through study enrollment.

MMSE, Mini-Mental State Examination; COWAT, controlled oral word association test.

Intra-dialytic Change in Blood Pressure

An increase in systolic blood pressure after the hemodialysis treatment was associated with steeper decline in executive function but not memory in unadjusted analyses. This association was attenuated and became non-significant after adjustment. Similarly, an increase in systolic blood pressure after hemodialysis was associated with steeper decline in three individual tests; however, after adjustment, only the association with the Word Recognition test remained significant. In quartile analyses, those participants with a rise in systolic blood pressure after hemodialysis were more likely to have executive function decline compared to those who experienced a fall in systolic blood pressure.

“U” shaped relationships

There was no evidence of a “U” shaped relationship between any of the blood pressure measures and decline in the composite memory or executive function score (All p values > 0.05)

Discussion

Lower pre-dialysis diastolic blood pressure and higher pulse pressure were both associated with steeper decline in cognitive function in maintenance hemodialysis patients. Systolic blood pressure, in contrast, showed no association with cognitive decline, indicating that low diastolic blood pressure is likely the component of pulse pressure that carries prognostic importance. The associations with decline were primarily seen with the executive function summary score and the component tests that comprise the executive function domain rather than with tests that assess memory. The observed difference in slopes leads to substantial changes in cognitive function over time. For those in the lowest diastolic quartile (or highest pulse pressure quartile), this translates to nearly a half standard deviation difference in test scores after five years. The difference in how fast participants with differing blood pressures decline appears significant viewed in comparison to the general population. For example, one such study in the general population reported a decline in MMSE score of 0.02 per year for participants aged 65 years and a decline in MMSE score of 0.57 per year for participants aged 85 years.²⁴ In comparison, individuals in this study with a pulse pressure in the highest quartile (mean difference in systolic and diastolic BP = 87 mmHg) experienced a decline in the MMSE of 0.48 per year greater than those in the lowest quartile (mean difference in systolic and diastolic BP = 50 mmHg).

In the general population, systolic and diastolic hypertension are associated with cognitive impairment², cognitive function decline³, and incident dementia,⁴ At least one study has also reported a non-linear, “U”-shaped relationship between blood pressure and cognitive function.²⁵ An analysis of the Baltimore Longitudinal Study of Aging demonstrated that both high and low diastolic blood pressures were associated with poorer executive function compared to normotensive individuals. It is less clear how these findings translate to those individuals with CKD, particularly those requiring maintenance hemodialysis, given the associations between low blood pressure and higher mortality in this population.^26–28

There is a substantial literature exploring the associations between blood pressure and mortality in patients receiving maintenance hemodialysis. An analysis of approximately 11,000 maintenance hemodialysis patients from the United States Renal Data System Waves 3 and 4 study demonstrated that low pre-dialysis systolic, low pre-dialysis diastolic, and high post-dialysis systolic blood pressure were all independently associated with all-cause mortality.²⁸ The authors concluded that widened pulse pressure was the strongest risk factor all-cause mortality. An observational study of over 5,000 hemodialysis patients by Zager et al noted a “U” shaped association between post dialysis systolic blood pressure and cardiovascular specific mortality²⁹ while Robinson et al found that in a cohort of over 4,000 maintenance hemodialysis patients that low pre-dialysis systolic blood pressure carried the greatest risk of mortality compared to other blood pressure measures.³⁰ Finally, a study by Kovesdy et al of more than 650,000 U.S. veterans with CKD found that those participants with low diastolic blood pressure (less than 70 mmHg) were at higher risk for all-cause mortality, even if the systolic blood pressure was considered “ideal”.³¹

Relatively few studies have investigated the link between blood pressure and cognitive function in patients with kidney disease. The Chronic Renal Insufficiency Cohort examined cross-sectional vascular risk factors for cognitive impairment in 3,591 participants with a mean eGFR of 43 ml/min/1.73m², finding that hypertension was a significant risk factor for cognitive impairment in unadjusted analyses but not after adjustment for kidney function. ⁹ A recent study by MacEwen et al in participants on maintenance hemodialysis investigated the interplay between blood pressure, cerebral ischemia (measured using cranial near-infrared spectroscopy), and cognitive function.³² Among the 58 participants, hypotension was associated with cerebral ischemia, but not cognitive impairment; however the small sample size was a limiting factor in drawing conclusions. In the same cohort of maintenance hemodialysis patients reported in this manuscript, we previously demonstrated an association between lower diastolic blood pressure and worse cognitive performance; however, we found that this relationship was attenuated and became non-significant when adjusting for demographics and comorbidity.¹² There are even less data on the impact of different blood pressure goals on cognitive function in CKD patients, and longitudinal data on cognitive function from the recently completed Systolic Blood Pressure Intervention Trial are eagerly awaited.³³

There are several potential mechanisms which could explain the association between lower blood pressure and cognitive decline in individuals with CKD. First, low blood pressure may be a marker of vascular disease. An analysis which combined data from the Cardiovascular Health Study and the Atherosclerosis Risk in Communities study demonstrated that low systolic blood pressure was a significant risk factor for stroke, but only in those participants with reduced kidney function.³⁴ Importantly, cerebrovascular disease most commonly produces a deficit within executive cognitive function, the type of impairment we found to be associated with lower blood pressure in this study. Second, those with low diastolic and high pulse pressure at the start of hemodialysis, a marker of loss of vascular elasticity³⁵, may be the most susceptible to intra-dialytic hypotension and impaired auto-regulation and thus decreased cerebral perfusion.³⁶ A recent study by Poliner-Bos et al examined the effect of hemodialysis on cerebral blood flow as measured by positron emission tomography-computed tomography imaging in twelve elderly participants.³⁷ Cerebral blood flow was found to decline by 10% during a hemodialysis treatment, suggesting that hemodialysis treatments may induce repetitive cerebral ischemic injury. Taken together, these lines of evidence may indicate that for patients receiving maintenance hemodialysis, standard blood pressure targets may not apply. Large scale randomized clinical trials, in which participants requiring hemodialysis are assigned to different blood pressure targets, may be the only definitive method to determine optimal blood pressure in this population.

The strengths of this study include use of an expansive neuro-cognitive battery, detailed ascertainment of potential confounding variables, and use of statistical modeling to account for the competing risk of death and kidney transplantation. There are also several limitations. The high mortality rate of our dialysis cohort limited the number of repeat cognitive measures, potentially reducing our ability to detect differences in cognitive decline; however, this mortality rate is consistent with the broader hemodialysis population and we accounted for the semi-competing risk of mortality (and the opposing risk of drop-out by transplantation) using joint models. We also included prevalent not incident hemodialysis patients, which may introduce selection bias to the cohort. Additionally, we excluded participants with either a reported history of dementia or a very low MMSE score (<10), which may underestimate the effect size associated with difference in blood pressure. Finally, our measure of blood pressure was obtained in the dialysis unit as part of standard clinical care, not via a standardized protocol, such as is done in many clinical trials. As a result, variation in cuff size selection, participant positioning, and timing may contribute to inaccurate blood pressure readings. Although we believe our use of the monthly mean blood pressure reduces the impact of day to day blood pressure variability, there is some evidence that home monitoring of blood pressure in dialysis patients may provide a more accurate measure than in-center measures.³⁸

Low diastolic blood pressure and widened pulse pressure were both shown to be risk factors for cognitive decline in a cohort of maintenance hemodialysis patients. Declines were most prominent in tests of executive function, suggesting that the association of diastolic blood pressure and pulse pressure may be mediated through cerebrovascular disease. Additional research should focus on determining optimal diastolic blood pressure targets in hemodialysis patients. Future studies are needed to determine if these findings extend to patients with CKD who do not yet require hemodialysis.

Table 5.

Association between intra-dialytic systolic BP change (per 10 mmHg higher) and difference in slope of cognitive tests (per year change)

Type of Test	Cognitive test	Scoring	n	N	Unadjusted Slope^	p	Adjusted Slope^ (95% CI)	p
Summary Score	Memory	SD*	300	645	0.00 (−0.02, 0.01)	0.7	0.00 (−0.02, 0.01)	0.7
Summary Score	Executive Function	SD*	300	645	−0.02 (−0.03, 0.00)	0.04	−0.02 (−0.03, 0.00)	0.1
Screening	MMSE	No. correct	314	785	−0.01 (−0.08, 0.06)	0.8	−0.01 (−0.07, 0.06)	0.9
Primarily Memory	Short Delayed Recall	No. correct	312	779	−0.02 (−0.07, 0.03)	0.4	−0.02 (−0.06, 0.02)	0.3
	Delayed Recall	No. correct	312	774	−0.03 (−0.08, 0.02)	0.2	−0.03 (−0.07, 0.01)	0.2
	Word Recognition	No. correct	312	775	−0.05 (−0.10, −0.01)	0.03	−0.05 (−0.09, −0.01)	<0.01
Primarily Executive Function	Block Design	No. completed	311	752	−0.18 (−0.34, −0.02)	0.03	−0.15 (−0.30, 0.00)	0.1
	Digit Symbol Substitution	No. completed	296	676	−0.11 (−0.35, 0.12)	0.4	−0.07 (−0.29, 0.15)	0.5
	Digits Forward	No. correct	187	452	0.03 (−0.02, 0.07)	0.2	0.03 (−0.01, 0.07)	0.2
	Digits Backward	No. correct	186	451	0.00 (−0.04, 0.05)	>0.99	0.00 (−0.04, 0.05)	0.9
	Trail Making Part A	Time to completion (sec)	302	693	0.49 (−0.86, 1.83)	0.5	0.44 (−0.89, 1.77)	0.5
	Trail Making Part B	Time to ompletion (sec)	299	681	0.92 (−0.54, 2.38)	0.2	0.66 (−0.69, 2.01)	0.3
	Mental Alternations	No. correct	187	453	0.04 (−0.10, 0.18)	0.6	0.05 (−0.09, 0.18)	0.5
	COWAT (Animals)	No. correct	187	453	−0.12 (−0.23, −0.01)	0.04	−0.10 (−0.21, 0.01)	0.1
	COWAT (Supermarket)	No. correct	187	454	−0.07 (−0.20, 0.06)	0.3	−0.05 (−0.18, 0.08)	0.4

^{^}Slope is the rate of change in the cognitive test per year.

^*Summary scores represent a per standard deviation change, with a mean of zero and SD of 1.

Negative coefficients are associated with steeper decline in scores over time except on the Trail Making Tests.

Bold indicates statistically significant differences over time.

n = number of unique patients and N = number of follow up visits. The Digits, Mental Alternations, and COWAT tests were added halfway through study enrollment.

MMSE, Mini-Mental State Examination; COWAT, controlled oral word association test.