Visit-To-Visit Blood Pressure Variability and the Risk of Dementia in Older People

Abstract

Background:

High visit-to-visit variability (VVV) in blood pressure (BP) is associated with cerebrovascular lesions on neuroimaging.

Objective:

Our primary objective was to investigate whether VVV is associated with incident all-cause dementia. As a secondary objective, we studied the association of VVV with cognitive decline and cardiovascular disease (CVD).

Methods:

We included community-dwelling people (age 70–78 year) from the ‘Prevention of Dementia by Intensive Vascular Care’ (preDIVA) trial with three to five 2-yearly BP measurements during 6–8 years follow-up. VVV was defined using coefficient of variation (CV; SD/mean×100). Cognitive decline was assessed using the Mini-Mental State Examination (MMSE). Incident CVD was defined as myocardial infarction or stroke. We used a Cox proportional hazard regression and mixed-effects model adjusted for sociodemographic factors and cardiovascular risk factors.

Results:

In 2,305 participants (aged 74.2±2.5), mean systolic BP over all available visits was 150.1 mmHg (SD 13.6), yielding a CV of 9.0. After 6.4 years (SD 0.8) follow-up, 110 (4.8%) participants developed dementia and 140 (6.1%) CVD. Higher VVV was not associated with increased risk of dementia (hazard ratio [HR] 1.00 per point CV increase; 95% confidence interval [CI] 0.96–1.05), although the highest quartile of VVV was associated with stronger decline in MMSE (β –0.09, 95% CI –0.17 to –0.01). Higher VVV was associated with incident CVD (HR 1.07; 95% CI 1.04–1.11).

Conclusion:

In our study among older people, high VVV is not associated with incident all-cause dementia. It is associated with decline in MMSE and incident CVD.

Keywords

Blood pressure blood pressure variability cognition cardiovascular disease dementia

INTRODUCTION

Blood pressure (BP) and dementia are intricately linked. At older age, both a high systolic BP (SBP), low diastolic BP (DBP), and decreasing BP over time appear to be associated with an increased risk for dementia [1, 2]. Recent studies have suggested that a high visit-to-visit variability (VVV) in BP might also be associated with an increased risk of cognitive decline and dementia [3 –5]. BP variability can be measured over a short period (minutes to hours) or with longer intervals, i.e., visit-to-visit [3]. VVV is a relative new construct that is often available in clinical practice. It is associated with an increased risk of cardiovascular disease (CVD) and mortality possibly due to increased arterial stiffness and/or poor adherence to antihypertensive treatment causing both a high VVV and an increased risk of CVD [6]. The increased occurrence of cerebrovascular lesions on neuroimaging in persons with high VVV might contribute to decreased cognitive functioning and ultimately dementia [3 , 7]. The association between VVV and cognitive impairment has been established in several studies [4, 8]. Until now only one study assessed the association with incident dementia and found a positive association [5].

In this study, we investigated whether VVV is associated with an increased risk of all-cause dementia in community-dwelling older people. Secondarily we assessed the association of VVV with cognitive decline and CVD.

METHODS

Study population and procedures

We performed our post-hoc analyses on data from the ‘Prevention of Dementia by Intensive Vascular Care’ (preDIVA) trial (ISRCTN29711771) [9]. In short, this was a multisite, cluster-randomized, open-label trial on the effect of intensive vascular care on incident dementia. The intervention consisted of four-monthly visits to a practice nurse. During these visits, cardiovascular risk factors were assessed and lifestyle advice and drug treatment was provided conforming to national guidelines. The guidelines did not include management of VVV. Participants in the control condition received care as usual. For the current analyses, we considered the trial population as a single-cohort. All community-dwelling older people (aged 70 to 78 years) registered to the participating primary care practices, were invited to participate [10]. The only exclusion criteria were a diagnosis of dementia or a disorder likely to hinder successful long-term follow-up. Intervention and follow-up were 6–8 years. The medical ethics committee of the Academic Medical Center (AMC) in Amsterdam approved the study and all participants gave written informed consent.

At baseline and after two, four, and six years, all participants visited the study nurse for an in-person assessment during which BP and other values were measured. Participants recruited early in the trial had a fifth visit after 7-8 years follow-up [9]. Only participants who completed three or more visits were included in the current analyses, as BP measurements at three separate visits were deemed as a minimum requirement for reliable determination of VVV. The BP measurements were performed in sitting position, using an automated BP monitor (M6, OMRON Healthcare Co., Ltd., Kyoto, Japan) [11]. BP was measured twice at the same arm during each visit. For each visit, the mean BP of the two measurements was calculated and per visit one BP value, i.e., the mean, was used to calculate VVV. Data on demographic characteristics, cardiovascular history, diabetes mellitus, medication use, and smoking habits were self-reported and cross-referenced with electronic health records. Weight and length were measured in order to calculate body mass index (BMI), and a blood-sample was obtained for measurement of the low-density lipoprotein (LDL) cholesterol.

Definition of visit-to-visit blood pressure variability

There are several measures for VVV, including Standard Deviation (SD), Coefficient of Variation (CV; calculated as SD divided by mean BP over all available visits, times 100), variation independent of mean (VIM; a transformation of SD uncorrelated to mean BP), average real variability (ARV; the average of absolute differences between successive measurements) and delta BP (maximum BP minus minimum BP) [12]. The main difference between these measures is the extent to which they depend on mean and absolute BP, and the influence of order of the BP measurements. For our analysis, we deemed CV most appropriate as our primary VVV parameter, as it is independent from mean BP and allows our results to be compared to other studies, which mostly used CV [5, 13]. Secondary analyses were conducted using the other VVV measures. Our primary analyses were executed with VVV based on SBP, but in secondary analyses we used VVV based on DBP [5, 13]. To assess the influence of minimum and maximum SBP we divided participants in four mutually exclusive categories: stable normotension (maximum ≤140 mmHg), episodic moderate hypertension (minimum SBP ≤140 mmHg and maximum SBP 140–179 mmHg), episodic severe hypertension (minimum SBP ≤140 mmHg and maximum SBP ≥180 mmHg), and stable hypertension (all SBP >140 mmHg) [13].

Outcomes

Participants were referred to their general practitioner to evaluate the possibility of a cognitive disorder in case of cognitive complaints, a decline in Mini-Mental State Examination (MMSE) of ≥3 points since baseline or ≥2 point since the preceding visit, or with a MMSE of ≤24 [9]. In case participants dropped-out from the trial information was retrieved on dementia status by contacting the participant, a relative, or the general practitioner. The primary outcome measure all-cause dementia was diagnosed according to the criteria of the Diagnostic and Statistical Manual of Mental Disorders IV (DSM-IV) [14]. An independent outcome adjudication committee blinded to treatment allocation assessed all possible and probable dementia cases based on available clinical data. Diagnosis of dementia was re-evaluated after one year, to minimize the risk of false-positive diagnoses. The committee also classified dementia cases into Alzheimer’s disease, vascular dementia, and other dementia types. Cognitive functioning was measured at each visit using the MMSE [15]. CVD was defined as incident myocardial infarction or stroke, and included both morbidity and mortality. CVD morbidity was self-reported, cross-referenced with electronic health records, and CVD mortality was based on data from death certificates. In the first four years of follow-up, TIA and stroke were collected as a combined outcome. Stroke diagnosed in the first four years was therefore not included in the analyses.

Statistical analysis

We assessed the association between VVV and dementia using a Cox proportional hazards regression model. The number of days from baseline to date of diagnosis, final follow-up visit, or time of death was used as timescale. VVV was assessed both as continuous variable and divided into quartiles to investigate a possible non-linear relationship. We first assessed the unadjusted association (model 1). In a second model we adjusted for sex, age, and low educational level defined as no education or primary education only. In model three, we additionally adjusted for obesity, defined as BMI ≥30 kg/m², LDL cholesterol, smoking, and diabetes mellitus. These covariates were considered potential confounders as they are known to be independently associated with an increased risk of dementia [16]. Data is presented based on model three, unless indicated otherwise. The proportional hazards assumption was assessed visually using log-minus-log plots and Schoenfeld residuals [17]. We analyzed the relation between VVV and changes in MMSE over time with a linear mixed-effects model. For this analysis, we used VVV divided into quartiles and the lowest quartile was used as reference group. Different models were fitted, but a mixed model with the VVV quartiles, time (visits), and their interaction as fixed effects was deemed most appropriate as we were interested in the association between VVV and changes in cognition over time. To assess the association of VVV, as continuous variable and divided into quartiles, with CVD, we used a Cox proportional hazards regression model. Because mortality is a competing risk for dementia and is associated with a high VVV, it could lead to a type II error with respect to a possible association between VVV and dementia. Therefore, we conducted a competing risk analysis according to the cause-specific hazard method and calculated the subdistribution hazard ratio (HR) [18]. The following predefined subgroup analyses were performed, including intervention versus control group, antihypertensive medication use at baseline, change in antihypertensive medication use (i.e., started or stopped) during follow-up, history of CVD (including myocardial infarction and stroke) at baseline, and number of BP measurements. To assess the influence of selective drop-out on our results, we repeated our analyses with VVV only based on the first three BP measurements. To assess whether an association between VVV and dementia might be affected by overall changes in BP, we additionally adjusted for the slope in SBP over all available visits [19]. To assess if the results of our analyses are indeed independent from hypertension at baseline (defined as SBP >140 mmHg, DBP >90 mmHg and/or use of antihypertensive medication) or mean SBP during follow-up, we additionally adjusted for these variables. In previous reports, adherence to antihypertensive treatment and seasonal change were identified to potentially influence VVV [20]. The association between these variables and VVV was assessed with unpaired t tests. Adherence to antihypertensive treatment was operationalized as self-reported full compliance with the prescribed medication regimen. Seasonal change was defined as at least two visits in different seasons. All statistical tests were two-sided with a p-value of <0.05 considered statistically significant. For the eight different VVV measures, a p-value of <0.006 was considered statistically significant after Bonferroni correction for multiple testing. Missing data were not imputed (see Supplementary Material 1, for additional information on statistical analyses). Analyses were done using the Statistical Package for Social Sciences version 24.0 (SPSS Inc., Chicago, IL, USA) and R studio version 3.3.3 [21].

RESULTS

From the 3,526 participants at baseline, 2,305 attended three or more visits and could be included in the analyses (Fig. 1, the compact version of the flow-diagram; Supplementary Figure 1, the complete flow-diagram). Baseline characteristics are shown in Table 1. Participants included in the analyses (i.e., those attending three to five visits) were slightly younger, had a lower SBP, less often a history of CVD, and had a higher MMSE at baseline compared to participants excluded from the analyses (Supplementary Table 1).

Fig.1

Flow-diagram of the number of participants in- and excluded from the analyses. Participants were included in the analyses if they had a valid blood pressure measurement at ≥3 visits. N, number of participants; VVV, visit-to-visit variability; MMSE, Mini-Mental State Examination; CVD, cardiovascular disease.

Table 1

Baseline characteristics of participants who did or did not develop dementia

	All participants		Dementia		No dementia		p value
	(n = 2305)		(n = 110)		(n = 2165)
Sex (male)	1032	(44.8%)	52	(47.3%)	58	(52.7%)	0.59
Age (y)	74.2	(2.5)	74.7	(2.5)	74.1	(2.4)	0.02
Ethnicity (Caucasian)	2191	(96.5%)	105	(96.3%)	2057	(96.5%)	0.86
Education (low)^*	493	(21.6%)	32	(29.9%)	453	(21.1%)	0.03
History of CVD	756	(33.0%)	44	(40.0%)	696	(32.4%)	0.10
Diabetes mellitus	437	(19.0%)	29	(26.4%)	403	(18.6%)	0.04
Antiglycemic med.	326	(14.1%)	21	(19.1%)	302	(13.9%)	0.17
Hypertension^**	2049	(88.9%)	97	(88.2%)	1923	(88.8%)	0.90
Antihypertensive med.	1246	(54.1%)	66	(60.0%)	1160	(53.6%)	0.19
Systolic BP (mmHg)	154.7	(20.8)	156.1	(21.5)	154.7	(20.8)	0.50
Diastolic BP (mmHg)	81.3	(10.6)	81.5	(11.2)	81.3	(10.6)	0.85
BMI (kg/m²)	27.5	(4.2)	27.6	(4.0)	27.5	(4.1)	0.71
LDL cholesterol (mmol/L)	3.1	(1.0)	3.0	(1.0)	3.1	(1.0)	0.07
Cholesterol lowering med.	774	(33.6%)	39	(35.5%)	728	(33.7%)	0.78
Current smoker	279	(12.1%)	14	(12.7%)	258	(11.9%)	0.81
MMSE	29	[28–30]	28	[26–29]	29	[28–30]	<0.01
Intervention group	1233	(53.5%)	43	(39.1%)	1010	(46.7%)	0.12

Baseline characteristics of all participants included in the analyses (i.e., present at ≥3 visits). Data are presented as frequencies (%), mean (SD) or median [interquartile range]. P values are calculated with a Chi-squared test, an unpaired T test or Mann-Whitney U test. ^*Low educational level defined as no education or primary education only. ^**Hypertension was defined as a systolic blood pressure at baseline ≥140 mmHg, a diastolic blood pressure at baseline ≥90 mmHg or the use of antihypertensive medication. CVD, cardiovascular disease; med., medication; BP, blood pressure; BMI, body mass index; LDL, low-density lipoprotein; MMSE, Mini-Mental State Examination.

Mean SBP was 154.7 mmHg (SD 20.8) at baseline and 150.6 mmHg (SD 20.9) at the final follow-up visit (Supplementary Figure 2). Mean DBP was 81.3 mmHg (SD 10.6) at baseline and 77.3 mmHg (SD 11.3) at final follow-up. Mean CV of SBP was 9.0 (13.6 [SD] / 150.1 [mean]×100) (Supplementary Table 2). As an example of the level of variability in SBP over time, we show the course in absolute SBP of four arbitrarily chosen participants representing the four quartiles of VVV (Supplementary Figure 3).

After an average follow-up of 6.4 years (SD 0.8), 110 participants (4.8%) were diagnosed with dementia (incidence rate: 7.0 per 1000 person-years). Of these, 82 (75%) had Alzheimer’s disease, six (5%) vascular dementia, five (5%) dementia from another etiology, and in 17 participants the type of dementia could not be classified. A higher CV of SBP was not associated with an increased risk of all-cause dementia (HR 1.00 per one point increase in CV of SBP, 95% confidence interval [CI] 0.96–1.05) (Table 2). No association with dementia risk and other VVV parameters were found (Supplementary Table 2). Also when stratified according to different BP categories, there was no association with dementia (Supplementary Table 3). VVV divided into quartiles did also not reveal a non-linear association with dementia (Fig. 2a). The number of dementia cases was 30 (5.5%) in the lowest quartile of VVV, 19 (3.3%, HR 0.6, 95% CI 0.3–1.1) in the second quartile, 32 (5.5%, HR 1.0, 95% CI 0.6–1.6) in the third quartile, and 29 (5.0%, HR 0.9, 95% CI 0.5–1.5) in the highest quartile. A higher CV of SBP was not associated with Alzheimer’s disease (HR 1.00, 95% 0.96–1.06).

The MMSE of participants in the highest quartile of VVV (range CV of SBP 11.3–36.0) was 0.18 (95% CI 0.00 to 0.36) points lower at baseline and declined on average with 0.09 (95% CI 0.01 to 0.17) points per visit more than participants in the lowest quartile of VVV (range CV of SBP 0.4–5.9) (Supplementary Table 4). During follow-up CVD occurred in 140 participants (6.1%), of which 93 had a myocardial infarction and 47 a stroke. A higher CV of SBP was significantly associated with more incident CVD (HR 1.07; 95% CI 1.04–1.11) (Table 2) and VVV divided into quartiles showed a linear association (Fig. 2b). Separate analyses yielded comparable results for myocardial infarction (HR 1.07; 95% CI 1.03–1.12) and stroke (HR 1.08; 95% CI 1.03–1.14).

Sensitivity analysis showed that there was no association between VVV and dementia when taking mortality into account in the competing risk analysis (subdistribution HR 1.01; 95% CI 0.96–1.05) (Supplementary Table 5). The predefined subgroup analyses yielded no apparent associations either (Table 3). No association with dementia was apparent when only including the first three BP measurements in the VVV measure (HR 1.01, 95% CI 0.97–1.05). Additional adjustment for trend in BP (HR 0.95, 95% CI 0.90–1.00, p = 0.06), hypertension at baseline (HR 1.01, 95% CI 0.96–1.05) or mean SBP during follow-up (HR 1.01, 95% CI 0.96–1.05) did not change the association with dementia. Adherence to antihypertensive treatment did not significantly influence CV of SBP (adherent, n = 1,563, mean 9.3, SD 4.6; non-adherent, n = 92, mean 9.2, SD 4.6; p = 0.83), nor did seasonal change (BP measured in different seasons, n = 615, mean 9.0, SD 4.4; BP measured in the same season, n = 1,690, mean 9.1, SD 4.5; p = 0.64).

Table 2

Cox proportional hazards regression on the association between blood pressure variability and dementia or cardiovascular disease

	Dementia			Cardiovascular disease
	Hazard ratio (95% CI)		p value	Hazard ratio (95% CI)		p value
Model 1	1.00	(0.96–1.04)	0.97	1.08	(1.04–1.11)	<0.01
Model 2	1.00	(0.96–1.05)	0.90	1.08	(1.04–1.11)	<0.01
Model 3	1.00	(0.96–1.05)	0.58	1.07	(1.04–1.11)	<0.01

Cardiovascular disease includes myocardial infarction or stroke. Model 1 is the unadjusted model; in model 2 we adjusted for gender, age, and low educational level; and in model 3 we additionally adjusted for obesity, low-density-lipoprotein, smoking, and diabetes. CI, confidence interval.

Fig.2

Hazard ratios of the association between quartiles of coefficient of variation of systolic blood pressure and dementia (A) or cardiovascular disease (B). Analyses are unadjusted. The y-axis is shown in logarithmic scale. Quartile 1 includes CV of SBP 0.4 to 5.9; quartile 2, CV of SBP 5.9 to 8.4; quartile 3, CV of SBP 8.4 to 11.3; quartile 4, CV of SBP 11.3 to 36.0. At the bottom of each graph, the number of participants with dementia or cardiovascular disease are shown in relation to the total number of participants per quartile. CI, confidence interval; VVV, visit-to-visit variability; CV, coefficient of variation; SBP, systolic blood pressure; HR, hazard ratio; ref., reference group.

DISCUSSION

In our study population of community-dwelling older people, VVV was not associated with incident, all-cause dementia after an average follow-up of 6.4 years. This result is irrespective of the measure of VVV applied, mean SBP, use of antihypertensive medication, or history of CVD. High VVV was associated with stronger decline in MMSE and with a higher incidence of CVD. The absence of an association with dementia is in contrast with findings from the Three-City Study, in which a significant association between a higher VVV and an increased risk of incident dementia was found [5]. This observational cohort study (n = 6,506 participants) is comparable regarding age, but had a lower systolic VVV (CV of SBP, 7.2) and cardiovascular risk at baseline in comparison to our study population, with a lower mean SBP and fewer participants with diabetes mellitus. The calculation of VVV and the follow-up duration was comparable to our analyses. In the Three-City study a higher incidence rate was found of 11.8 dementia cases per 1000 person-years, possibly due to a more sensitive dementia assessment procedure using a neuropsychological test battery as opposed to the pragmatic and clinical approach used in preDIVA. However, it seems unlikely that this influenced the association with VVV. In a subgroup of our study population without a history of CVD, there was a trend toward a positive association between a higher VVV and incident dementia, although this effect was small. Perhaps in older people with a higher cardiovascular burden, cerebrovascular damage, presumed to contribute to the occurrence of cognitive decline and dementia, is already too advanced to detect a significant influence of VVV on incident dementia [5, 7]. Another theoretical possibility explaining the contrast between our findings and those of the Three-City study is that our study was underpowered to detect an association. This seems unlikely since the HR for dementia was one; not suggestive of a small sample size as a cause of our null finding. Reproduction of these analyses in different study populations may be required to determine the true nature of the association between VVV and dementia incidence.

Table 3

Subgroup analyses of the association between blood pressure variability and dementia

	Dementia cases		CV		HR	p value
	(n=, %)		(mean, SD)		(95% CI)
Intervention group	67 / 1222	(5.5%)	9.3	(4.5)	0.98 (0.91–1.06)	0.66
Control group	43 / 1053	(4.1%)	8.8	(4.3)	1.01 (0.96–1.06)	0.81
AHM at baseline	66 / 1160	(5.4%)	9.4	(4.5)	1.04 (0.97–1.11)	0.30
No AHM at baseline	44 / 1047	(4.2%)	8.6	(4.3)	0.97 (0.92–1.03)	0.36
AHM started/stopped^*	32 / 579	(5.5%)	9.7	(4.7)	0.99 (0.94–1.04)	0.70
No AHM started/stopped^**	78 / 1696	(4.6%)	8.8	(4.3)	1.02 (0.95–1.10)	0.58
History of CVD	44 / 740	(5.9%)	9.4	(4.8)	0.93 (0.86–1.00)	0.05
No history of CVD	66 / 1520	(4.3%)	8.9	(4.3)	1.04 (0.99–1.10)	0.11
3 BP measurements	50 / 455	(11.0%)	8.4	(4.7)	1.01 (0.96–1.07)	0.86
4 BP measurements	43 / 999	(4.3%)	9.1	(4.7)	1.01 (0.95–1.07)	0.82
5 BP measurements	17 / 851	(2.0%)	9.4	(4.0)	1.06 (0.95–1.18)	0.35

Analyses are unadjusted (model 1). ^*Antihypertensive medication started or stopped during follow-up. ^**Antihypertensive medication (yes or no) constant throughout study. Only the interaction term of ‘history of CVD’ and CV was significant (p = 0.01). CV, coefficient of variation; SD, standard deviation; AHM, antihypertensive medication; CVD, cardiovascular disease (including myocardial infarction and stroke); CI, confidence interval.

One of the hypotheses on a potential association between VVV and dementia is through progression of cerebrovascular lesions including white matter hyperintensities, cortical infarcts, and cerebral microbleeds [4, 7]. We found an association between VVV and increased risk of stroke, and between VVV and decline in MMSE. It is conceivable that the 6–8 years of follow-up in our study was too short to detect a clinically overt effect on dementia incidence as a result of high VVV, or that at the age range in our study cerebrovascular damage has advanced too much for VVV to influence the progression. The relation between high BPV earlier in life (i.e., midlife or early late-life) and dementia in late life may be stronger, similar to the relation between absolute blood pressure in mid-life and dementia in late-life [1]. Other potential mechanisms underlying an association with dementia are that high VVV and cerebral amyloid-β depositions (an important neuropathological hallmark of Alzheimer’s disease) are the result of increased arterial stiffness [22]. The neuropathological changes characteristic for Alzheimer’s disease might also lead to autonomic dysfunction and through this a higher VVV, even before any cognitive problems have occurred [8]. Finally, a potential relation between VVV and dementia might not be causal, but both stem from a common underlying cause.

We found a significant association between VVV and decline in MMSE. This is in accordance with previous reports that found a positive association between VVV and cognitive deterioration [4 , 23]. However, the association found in our study is probably not clinically relevant, as it would take approximately 18 years for the MMSE to be one point lower in the highest versus the lowest VVV quartile. The MMSE is not very sensitive to minor cognitive changes and was developed as a screening tool for dementia [15].

The association of VVV with CVD may have consequences for clinical practice. Potentially, antihypertensive treatment should not only be initiated and evaluated to reduce mean BP, but also to reduce VVV. The effect of different antihypertensive drugs on VVV is variable, and the lowest VVV is seen in treatment with calcium-channel blockers and non-loop diuretics [24]. The drug-class differences can partially account for the difference in stroke risk, but not for the difference in risk of myocardial infarction [24]. Interestingly, calcium channel blockers belong to the class of antihypertensive drugs associated with a lower risk of dementia [25].

Strengths of our study are the large number of participants and virtually complete information on primary outcome. The assessment of dementia was thorough with a long follow-up and validation by an independent outcome adjudication committee, since it concerned the primary outcome of the preDIVA study [9]. We added several sensitivity analyses, including other VVV measures, to strengthen our findings. An important limitation of our study is that the number of available BP measurements is limited and the interval between BP measurements is relatively long. Analyses stratified for number of BP measurements showed that participants with only three BP measurements had on average a lower VVV (CV 8.4) and a non-significantly higher risk of dementia (11%). To assess whether selective drop-out influenced our results, we repeated our analyses with VVV based only on the first three BP measurements, but this did not change the results. Some participants who were excluded from the analyses because of <3 BP measurements had an MMSE <24 at their last visit. Due to the limited number of BP measurements, we are unable to assess whether these participants also had a relative high VVV. Another limitation of our analyses is that they are based on a randomized controlled trial population. The intervention may have influenced variability, particularly early on in the study when hypertension was newly diagnosed. However, in our stratified analysis the influence of randomization group appeared minimal and the difference in BP reduction between intervention and control groups was small. We did not collect data on dosage of antihypertensive medication and time of day of the BP measurement. We could therefore not assess the influence of these factors on VVV. We were, however, able to assess the influence of adherence to antihypertensive treatment and seasonal change. Stroke occurrence is possibly underreported in our study due to the fact that it was not collected separately from TIA during the early phases of the study, and incident cases in the first four years of follow-up were therefore not included in the analyses. However, these events were missing for all participants and it is unlikely that this influenced the association with VVV.

In conclusion, in our study population of community-dwelling older people VVV is not associated with an increased risk of dementia over six years of follow-up. Future research is needed to confirm the findings of this study and assess whether other associations between VVV and incident dementia might exist including at a younger age and if followed over longer periods. We found a significant association with decline in MMSE, but its clinical relevance is uncertain. High VVV was associated with more incident CVD.

Footnotes

ACKNOWLEDGMENTS

We would like to thank all participants, practice nurses, and general practitioners involved in the preDIVA study. We also would like to thank Lisa S.M. Eurelings, Suzanne A. Ligthart, Carin E. Miedema, and Marieke P. Hoevenaar-Blom for their hard work for the preDIVA study. The preDIVA trial was supported by the Dutch Ministry of Health, Welfare and Sports (50-50110-98-020), the Innovatiefonds Zorgverzekeraars (innovation fund of collaborative health insurances, 05-234), and ZonMw (The Netherlands Organisation for Health Research and Development, 62000015).

The analyses presented here are based on funding from the European Union Seventh Framework Programme (FP7/2007-2013) under grant agreement n° 305374 and an EU Joint Programme - Neurodegenerative Disease Research (JPND) project. The JPND project is supported through the following funding organisations under the aegis of JPND – : Finland, Suomen Akatemia (Academy of Finland,291803); France, L’Agence Nationale de la Recherche (The French National Research Agency, ANR-14-JPPS-0001-02); Germany, Bundesministerium für Bildung und Forschung (BMBF) (The Federal Ministry of Education and Research, FKZ01ED1509); Sweden, Vetenskapsrådet (VR) (Swedish Research Council, 529-2014-7503), The Netherlands, ZonMw (The Netherlands Organisation for Health Research and Development, 733051041).

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Qiu

, Winblad

, Fratiglioni

(2005) The age-dependent relation of blood pressure to cognitive function and dementia. Lancet Neurol 4, 487–499.

Stewart

, Xue

, Masaki

, Petrovitch

, Ross

, White

, Launer

(2009) Change in blood pressure and incident dementia: A 32-year prospective study. Hypertension 54, 233–240.

Rothwell

(2010) Limitations of the usual blood-pressure hypothesis and importance of variability, instability, and episodic hypertension. Lancet 375, 938–948.

Sabayan

, Wijsman

, Foster-Dingley

, Stott

, Ford

, Buckley

, Sattar

, Jukema

, van Osch

, van der Grond

, van Buchem

, Westendorp

, de Craen

, Mooijaart

(2013) Association of visit-to-visit variability in blood pressure with cognitive function in old age: Prospective cohort study. BMJ 347, f4600.

Alperovitch

, Blachier

, Soumare

, Ritchie

, Dartigues

, Richard-Harston

, Tzourio

(2014) Blood pressure variability and risk of dementia in an elderly cohort, the Three-City Study. Alzheimers Dement 10, 330S337.

Muntner

, Whittle

, Lynch

, Colantonio

, Simpson

, Einhorn

, Levitan

, Whelton

, Cushman

, Louis

, Davis

, Oparil

(2015) Visit-to-visit variability of blood pressure and coronary heart disease, stroke, heart failure, and mortality: A cohort study. Ann Intern Med 163, 329–338.

Brickman

, Reitz

, Luchsinger

, Manly

, Schupf

, Muraskin

, DeCarli

, Brown

, Mayeux

(2010) Long-term blood pressure fluctuation and cerebrovascular disease in an elderly cohort. Arch Neurol 67, 564–569.

Nagai

, Dote

, Kato

, Sasaki

, Oda

, Kagawa

, Nakano

, Yamane

, Higashihara

, Miyauchi

, Tsuchiya

(2017) Visit-to-visit blood pressure variability and Alzheimer’s disease: Links and risks. J Alzheimers Dis 59, 515–526.

Moll van Charante

, Richard

, Eurelings

, van Dalen

, Ligthart

, van Bussel

, Hoevenaar-Blom

, Vermeulen

, van Gool

(2016) Effectiveness of a 6-year multidomain vascular care intervention to prevent dementia (preDIVA): A cluster-randomised controlled trial. Lancet 388, 797–805.

10.

Richard

, Van den Heuvel

, Moll van Charante

, Achthoven

, Vermeulen

, Bindels

, Van Gool

(2009) Prevention of dementia by intensive vascular care (PreDIVA): A cluster-randomized trial in progress. Alzheimer Dis Assoc Disord 23, 198–204.

11.

Altunkan

, Iliman

, Altunkan

(2008) Validation of the Omron M6 (HEM-7001-E) upper arm blood pressure measuring device according to the International Protocol in elderly patients. Blood Press Monit 13, 117–122.

12.

Webb

, Fischer

, Rothwell

(2011) Effects of beta-blocker selectivity on blood pressure variability and stroke: A systematic review. Neurology 77, 731–737.

13.

Rothwell

, Howard

, Dolan

, O’Brien

, Dobson

, Dahlof

, Sever

, Poulter

(2010) Prognostic significance of visit-to-visit variability, maximum systolic blood pressure, and episodic hypertension. Lancet 375, 895–905.

14.

American Psychiatric Association (2000), Diagnostic and Statistical Manual of Mental Disorders, Washington, DC.

15.

Folstein

, Folstein

, McHugh

(1975) “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 12, 189–198.

16.

Norton

, Matthews

, Barnes

, Yaffe

, Brayne

(2014) Potential for primary prevention of Alzheimer’s disease: An analysis of population-based data. Lancet Neurol 13, 788–794.

17.

Bellera

, MacGrogan

, Debled

, de Lara

, Brouste

, Mathoulin-Pelissier

(2010) Variables with time-varying effects and the Cox model: Some statistical concepts illustrated with a prognostic factor study in breast cancer. BMC Med Res Methodol 10, 20.

18.

Hinchliffe

, Lambert

(2013) Flexible parametric modelling of cause-specific hazards to estimate cumulative incidence functions. BMC Med Res Methodol 13, 13.

19.

Poortvliet

RKE

, de Ruijter

, de Craen

AJM

, Mooijaart

, Westendorp

RGJ

, Assendelft

WJJ

, Gussekloo

, Blom

(2013) Blood pressure trends and mortality: The Leiden 85-plus Study. J Hypertens 31, 63–70.

20.

Parati

, Ochoa

, Lombardi

, Bilo

(2013) Assessment and management of blood-pressure variability. Nat Rev Cardiol 10, 143–155.

21.

RStudio Team (2016) RStudio: Integrated Development for R. RStudio, Inc., Boston, MA. http://www.rstudio.com/.

22.

Hughes

, Kuller

, Barinas-Mitchell

, Mackey

, McDade

, Klunk

, Aizenstein

, Cohen

, Snitz

, Mathis

, Dekosky

, Lopez

(2013) Pulse wave velocity is associated with beta-amyloid deposition in the brains of very elderly adults. Neurology 81, 1711–1718.

23.

Nagai

, Hoshide

, Ishikawa

, Shimada

, Kario

(2012) Visit-to-visit blood pressure variations: New independent determinants for cognitive function in the elderly at high risk of cardiovascular disease. J Hypertens 30, 1556–1563.

24.

Webb

, Fischer

, Mehta

, Rothwell

(2010) Effects of antihypertensive-drug class on interindividual variation in blood pressure and risk of stroke: A systematic review and meta-analysis. Lancet 375, 906–915.

25.

van Middelaar

, van Vught