Long-term Blood Pressure Variability and Risk of Cardiovascular Disease Events Among Community-Dwelling Elderly

Abstract

High office blood pressure variability (OBPV) in midlife increases the risk of cardiovascular disease (CVD), but the impact of OBPV in older adults without previous CVD is unknown. We conducted a post-hoc analysis of ASPREE (ASPirin in Reducing Events in the Elderly) trial participants aged 70-years and older (65 for US minorities) without history of CVD events at baseline, to examine risk of incident CVD associated with long-term, visit-to-visit OBPV. CVD was a prespecified, adjudicated secondary endpoint in ASPREE. We estimated OBPV using within-individual standard deviation of mean systolic blood pressures (BP) from baseline and first two annual visits. Cox proportional hazards regression was used to calculate hazard ratios (HR) and 95% confidence intervals (CI) for associations with CVD events. In 16,475 participants who survived to year two without events, those in the highest tertile of OBPV had increased risk of CVD events after adjustment for multiple covariates, when compared to participants in the lowest tertile (HR=1.36; 95% CI: 1.08-1.70; P=0.01). Similar increased risk was observed for ischemic stroke (HR=1.56; 95% CI: 1.04-2.33; P=0.03), heart failure hospitalization or death (HR=1.73; 95% CI: 1.07-2.79; P=0.02), and all-cause mortality (HR=1.27; 95% CI: 1.04-1.54; P=0.02). Findings were consistent when stratifying participants by use of antihypertensive drugs, while sensitivity analyses suggested the increased risk was especially for individuals whose BP was uncontrolled during the OBPV estimation period. Our findings support increased OBPV as a risk factor for CVD events in healthy older adults with, or without hypertension, who have not had such events previously.

Clinicaltrials.gov registration: NCT01038583; ISRCTN.com registration: ISRCTN83772183

Graphical Abstract

Introduction

Blood pressure (BP) is a dynamic physiologic parameter regulated by distinct biological processes and influenced by intrinsic factors, including pserfusion demands by different organs, and extrinsic stimuli.¹ Short-term oscillations in BP (hourly, through twenty-four hours) can be characterized via ambulatory BP monitoring, while long-term fluctuation is ascertained through sequential home or office BP readings obtained longitudinally over days, weeks, months, and even years. The degree of change in amplitude, or variability of BP, occurring during these various time periods is increasingly recognized to demonstrate an independent relationship with organ damage which persists even after adjustment for average BP levels.^1-3

Long-term, visit-to-visit office blood pressure variability (OBPV) observed during midlife predicts late life cardiovascular events,^4-6 but the impact of long-term OBPV in a general population of adults who have survived into their seventh, eighth and ninth decades of life free of cardiovascular disease (CVD) events and disabling illnesses, is less certain. Previous studies examining the impact of OBPV on the risk of CVD have typically been conducted within middle-aged hypertensive cohorts or within BP intervention-based studies of high-risk individuals,^7-9 which lack generalizability to normotensive and hypertensive, but otherwise healthy, older adults. To address this gap, we examined the risk of incident CVD events associated with long-term, visit-to-visit OBPV in participants of the ASPirin in Reducing Events in the Elderly (ASPREE) study, a randomized primary prevention trial of daily low-dose aspirin conducted in non-disabled, community-dwelling older adults in Australia and the United States (US) who were in good health, and without prior CVD events at enrollment.¹⁰

Methods

The data (version 3.0) that support the findings of this study are available from the ASPREE Data Coordinating Center, Monash University School of Public Health (Aspree.AMS@monash.edu) upon reasonable request.

Study Participants

ASPREE was an international, multicenter, randomized, blinded outcome trial co-sponsored by the National Institute on Aging, National Cancer Institute, National Health and Medical Research Council (Australia), and other funding partners. The detailed methods of ASPREE, its recruitment and primary outcomes have been previously described.^11,12 In brief, ASPREE recruited community-dwelling adults from Australia and the US aged 70 years and older (65 years and older if US minority) from 2010-2014, who were free of documented evidence of CVD, dementia, physical disability, or any other chronic illness expected to limit survival to less than 5 years. Individuals with uncontrolled high BP (systolic BP ≥ 180 mmHg and/or diastolic ≥ 105 mmHg) were excluded from enrollment.

Enrolled subjects were randomized to receive daily low-dose aspirin 100 mg or placebo, and followed for a median of 4.7 years. Bayer AG provided aspirin and matching placebo, but otherwise had no role in the conduct of the trial. The trial adhered to the principles of the Declaration of Helsinki and international standards pertaining to protection of human subjects, with approval by local institutional review boards at each site. All participants provided written informed consent for their participation. The trial was registered at https://www.clinicaltrials.gov (NCT01038583) and https://www.isrctn.com (ISRCTN83772183).

Study Measurements

All participants enrolled into ASPREE completed two baseline visits to finalize eligibility, and after randomization were seen annually for study assessments conducted by trained study staff following standardized operating procedures. At these visits, participants underwent comprehensive evaluations of cognitive and physical function, as well as collection of anthropometric and laboratory measurements, medical comorbidities, lifestyle and socio-demographic factors, concomitant prescription medications, and other related health parameters.

Blood pressure for each participant was measured according to American Heart Association guidelines in the seated position after at least five minutes of rest using an automated oscillometric device with an occluding cuff of appropriate size for the upper arm circumference.¹³ The ASPREE protocol did not require a specific device model since it was not designed as a BP intervention study, but it did require use of a validated oscillometric device (OMRON HEM-7203, 7121, and 7130 models were exclusively used in Australia; Table S1 in the Data Supplement lists the devices used in the US sites). The study protocol specified that three separate and consecutive BP readings one minute apart be performed at each visit, and the mean of these measurements was recorded as the BP for the study visit. All study sites were regularly monitored to ensure compliance with the study protocol.

Assessment of Visit-to-Visit OBPV

To be included in this analysis, participants must have had at a mean BP value recorded for baseline, first and second year annual follow-up visits. We then estimated the OBPV for an individual using the standard deviation (SD) of the mean systolic BPs obtained from these three study visits. In addition to SD, we also calculated the coefficient of variation (CV), average real variability (ARV), and variation independent of the mean (VIM), as alternate estimates of OBPV.¹ We observed high correlations of these different indices with SD in our sample (Table S2); therefore, consistent with other studies examining long-term, visit-to-visit OBPV,^7-9 we used SD in our primary analysis.

Outcomes

The primary outcome of the ASPREE trial was disability-free survival, and these results have been previously reported.¹² Incident CVD was a pre-specified, composite secondary endpoint of ASPREE, consisting of fatal coronary heart disease (death from myocardial infarction, sudden cardiac death, cardiac failure death, or any other death in which the underlying cause was considered to be coronary heart disease), nonfatal myocardial infarction, fatal or nonfatal stroke (hemorrhagic or ischemic), non-coronary cardiac or vascular death, or hospitalization for heart failure.¹⁴ All CVD events were adjudicated as part of the main trial by a panel of experts blinded to treatment group assignments.

In this post-hoc analysis, we used the ASPREE-defined CVD composite endpoint¹⁴ as our primary outcome for exploring the association of OBPV with incident CVD events. In addition to the CVD composite endpoint, we also examined the association of OBPV separately for key outcomes strongly affected by BP, including ischemic stroke (among N=16,608), heart failure hospitalization or death (among N=16,641), and all-cause mortality (among N= 16,757), after excluding those who experienced the specific event of interest prior to the second annual visit, and after recalculating tertiles accordingly.

Statistical Analysis

Statistical analyses were performed using Stata version 15.1 for Windows (StataCorp LP, College Station, Texas, USA). Cox proportional hazards regression was used to calculate hazard ratios (HR) and 95% confidence intervals (CI) for time to the first event within each endpoint definition occurring during the follow-up period after the OBPV estimation period. These hazard ratios were cause-specific with participants censored at the time of the event of interest. We explored the relationship between OBPV and incident CVD, initially with OBPV as a continuous variable. Thereafter, we divided OBPV according to tertiles of SD, and used the lowest (T1) tertile as referent. The unadjusted Cox proportional-hazard model (Model 1) was followed with adjustment for accumulating sets of covariates, which included key demographic characteristics (Model 2), mean systolic BP during the OBPV estimation period (Model 3), and use of antihypertensive drugs at baseline (Model 4). All analyses were repeated in the same manner for the individual endpoints of ischemic stroke, hospitalization or death from heart failure, and all-cause mortality. In addition, cumulative incidence of the CVD endpoint as well as the other individual endpoints were computed according to the complement of the Kaplan–Meier estimates of event-free survival. Cumulative incidence functions were computed for individual endpoints to account for competing risks.

To assess the robustness of our results, we performed sensitivity analyses. First, we conducted an analysis stratifying participants based on in-trial BP control (<140/90 mmHg or >140/90 mmHg) during the OBPV estimation period. Secondly, we conducted an analysis stratifying participants based on use, or nonuse, of antihypertensive drugs at baseline. Thirdly, we repeated our primary analyses using ARV as the index of OBPV because it showed the least correlation with SD than the other indices of OBPV. Finally, we conducted a sensitivity analysis using an alternate criteria of BP measurements from four visits (baseline, Years 1, 2, and 3) to estimate OBPV, which then shifted Year 3 to the new baseline in the time-to-event analysis (and after excluding individuals with events prior to Year 3). In each of these analyses, we assessed the association of OBPV both as a continuous variable and divided into tertiles which were re-calculated accordingly.

Results

Of the original 19,114 individuals enrolled into ASPREE, 16,758 survived and remained in the study to the second annual visit and had mean BP recorded at baseline, first and second year annual visits, allowing for estimation of OBPV (Figure 1). From baseline to the second annual visit, 283 individuals experienced the CVD endpoint, leaving 16,475 participants available for the analysis of OBPV with the primary CVD endpoint. Table 1 shows the baseline characteristics of those 16,475 participants overall and according to SD tertiles of OBPV. From the lowest to highest tertile, participants were progressively older and had higher body mass index (BMI), higher baseline and average systolic BP, and higher rates of comorbidities such as hypertension, diabetes, and chronic kidney disease. We observed a weak correlation of OBPV with both baseline and average systolic BP (Figure S1).

Figure 1:

Office blood pressure variability estimation and outcome ascertainment timeline among the ASPREE participants. AV = annual visit; BPV = systolic blood pressure variability

Table 1.

Baseline characteristics of ASPirin in Reducing Events in the Elderly trial participants overall, and by tertiles of office systolic blood pressure variability (OBPV), who remained free of cardiovascular disease events through the second annual visit.

Characteristic N	Overall 16,475	T1 5,501	T2 5,494	T3 5,480	P-value for trend
Within individual SD of OBPV (mmHg)	10.1±6.0	4.3±1.7	9.2±1.4	17.0±4.7	<0.001
Baseline systolic BP (mmHg)	139±16	136±14	138±15	143±18	<0.001
Mean systolic BP (mmHg) through Year 2 annual visit	138±14	135±14	137±13	141±13	<0.001
Age, years (mean ± SD)	75.0±4.4	74.6±4.2	74.9±4.3	75.5±4.7	<0.001
Male (%)	43.9	46.0	44.1	41.7	<0.001
Race (%)
White	92.4	92.9	92.8	91.5	0.004
African-American	3.9	3.4	3.9	4.4	0.01
Hispanic	2.3	2.4	2.0	2.4	0.82
Asian	0.8	0.7	0.7	1.1	0.03
Other	0.6	0.5	0.6	0.7	0.25
BMI (kg/m2)	28.1±4.7	28.0±4.5	28.0±4.7	28.3±4.8	0.01
Education (years)
<12	44.9	45.3	44.2	45.0	0.76
12-15	28.8	27.7	29.5	29.2	0.10
16+	26.3	27.0	26.3	25.8	0.17
Comorbidities (%)
Hypertension	74.0	65.6	72.8	83.5	<0.001
Diabetes	10.3	9.6	9.8	11.6	<0.001
Smoking (ever)	43.9	44.1	43.1	44.5	0.67
Chronic Kidney Disease (eGFR <60 ml/min)	17.9	16.1	16.5	21.0	<0.001
Alcohol (current use)	77.6	77.7	78.1	76.9	0.34
Medication (%)
Statins	33.8	33.3	33.4	34.6	0.15
Antihypertensives	52.1	46.5	50.4	59.5	<0.001

T=tertile; SD = standard deviation; OBPV = office blood pressure variability; BP = blood pressure; BMI = body mass index

Following the second annual visit, 497 participants experienced one of the components of the composite CVD endpoint during a median follow-up period of 2.6 years (IQR=1.6 to 3.6 years) (Table 2). In univariate analyses (Model 1) examining OBPV as a continuous variable, we observed a significantly increased risk of CVD events which persisted when the analysis was subsequently adjusted for age, sex, diabetes, smoking, country, aspirin, ethnicity, BMI, statin use at baseline (Model 2), and further adjusted for within-individual mean systolic BP during the OBPV estimation period (Model 3), and use of antihypertensive drugs at baseline (Model 4). When OBPV was divided into tertiles, those participants in the highest tertile had significantly increased risk of CVD events compared to those in the lowest tertile across all models. In the fully adjusted model, the risk of incident CVD was increased by 36% for individuals in the highest tertile relative to the lowest tertile (HR=1.36; 95% CI: 1.08-1.70; P=0.01). A similar increased risk was observed for the middle tertile in relation to the lowest tertile, but the association was no longer statistically significant when adjusted for within-individual mean systolic BP and use of antihypertensive drugs.

Table 2.

Association of office blood pressure variability (OBPV) with incident cardiovascular disease events and other endpoints occurring after the second annual visit in ASPirin in Reducing Events in the Elderly trial participants.

Event Type	OBPV (continuous)		OBPV Tertiles
		T1	T2	T3
CVD, (N)	16,475	5,501	5,494	5,480
Number of events (%)	497 (3.0%)	124 (2.3%)	166 (3.0%)	207 (3.8%)
Event rate per 1000 person years	11.6 (10.6-12.7)	8.8 (7.4-10.5)	11.6 (9.9-13.5)	14.4 (12.6-16.5)
Hazard Ratio (95% CI)
*Model 1	1.04 (1.02-1.05) P<0.001	1.00	1.32 (1.04-1.66) P=0.02	1.64 (1.31-2.05) P<0.001
Ɨ Model 2	1.03 (1.02-1.04) P<0.001	1.00	1.27 (1.01-1.61) P=0.04	1.46 (1.17-1.83) P=0.001
ǂ Model 3	1.02 (1.01-1.04) P<0.001	1.00	1.25 (0.99-1.58) P=0.06	1.38 (1.10-1.73) P=0.005
§ Model 4	1.02 (1.01-1.04) p=0.001	1.00	1.24 (0.98-1.57) P=0.07	1.36 (1.08-1.70) P=0.01
Ischemic Stroke, (N)	16,608	5,567	5,533	5,508
Number of events (%)	164 (1.0%)	38 (0.7%)	57 (1.0%)	69 (1.3%)
Event rate per 1000 person years	3.8 (3.2-4.4)	2.6 (1.9-3.6)	3.9 (3.0-5.1)	4.7 (3.7-6.0)
Hazard Ratio (95% CI)
*Model 1	1.04 (1.02-1.07); P<0.001	1.00	1.47 (0.98-2.23); P=0.06	1.79 (1.21-2.66); P=0.004
Ɨ Model 2	1.04 (1.01-1.06); P=0.002	1.00	1.46 (0.97-2.21); P=0.07	1.70 (1.12-2.50); P=0.01
ǂModel 3	1.03 (1.01-1.05); P=0.017	1.00	1.42 (0.94-2.15); P=0.10	1.53 (1.02-2.29); P=0.04
§Model 4	1.03 (1.01-1.05); P=0.012	1.00	1.43 (0.94-2.16); P=0.09	1.56 (1.04-2.33); P=0.03
Heart Failure hospitalization or death, (N)	16,641	5,577	5,551	5,513
Number of events (%)	116 (0.7%)	25 (0.5%)	33 (0.6%)	58 (1.1%)
Event rate per 1000 person years	2.7 (2.2-3.2)	1.7 (1.2-2.6)	2.3 (1.6-3.2)	4.0 (3.1-5.1)
Hazard Ratio (95% CI)
*Model 1	1.06 (1.04-1.09); P<0.001	1.00	1.30 (0.77-2.19); P=0.32	2.30 (1.44-3.67); P=0.001
Ɨ Model 2	1.05 (1.02-1.07); P<0.001	1.00	1.22 (0.73-2.06); P=0.45	1.84 (1.14-2.96); P=0.01
ǂModel 3	1.04 (1.02-1.07); P=0.001	1.00	1.22 (0.72-2.05); P=0.46	1.78 (1.11-2.87); P=0.02
§Model 4	1.04 (1.01-1.07); P=0.002	1.00	1.21 (0.72-2.04); P=0.47	1.73 (1.07-2.79); P=0.02
All-cause Mortality (N)	16,757	5,610	5,581	5,566
Number of events (%)	584 (3.5%)	169 (3.0%)	159 (2.9%)	256 (4.6%)
Event rate per 1000 person years	13.0 (12.0-14.1)	11.4 (9.8-13.2)	10.6 (9.1-12.4)	16.9 (15.0-19.2)
Hazard Ratio (95% CI)
*Model 1	1.04 (1.03-1.05); P<0.001	1.00	0.92 (0.74-1.15); P=0.48	1.48 (1.22-1.79) P<0.001
Ɨ Model 2	1.03 (1.01-1.04); P<0.001	1.00	0.88 (0.71-1.09); P=0.24	1.25 (1.02-1.52) P=0.03
ǂModel 3	1.03 (1.02-1.04); P<0.001	1.00	0.88 (0.71-1.10); P=0.26	1.27 (2.04-1.55) P=0.02
§Model 4	1.03 (1.01-1.04); P<0.001	1.00	0.88 (0.71-1.10); P=0.26	1.27 (1.04-1.54); P=0.02

OBPV = office blood pressure variability; T = tertile; CVD = cardiovascular disease; CI = confidence interval

^*Unadjusted

^ƗAdjusted for age, sex, diabetes, smoking, country, aspirin, ethnicity, body mass index, statin use as baseline

^ǂModel 2 plus adjusted for within-individual mean systolic blood pressure from baseline to second annual visit

^§Model 3 plus adjusted for use of antihypertensive drugs at baseline

Following the second annual visit, 164 ischemic strokes, 116 heart failure events (hospitalization or death), and 584 deaths occurred. We observed a consistently increased risk for each of these outcomes with higher OBPV, when considered as either per 1 mmHg unit increase in OBPV, or highest versus lowest tertiles of OBPV, and which persisted even after full covariate adjustment (Table 2). Figure 2 shows the cumulative incidence of CVD, ischemic stroke, hospitalization or death from heart failure, and all-cause mortality by tertiles of OBPV.

Figure 2.

Cumulative incidence of cardiovascular disease (CVD) events (Panel A), all-cause mortality (Panel B), ischemic stroke (Panel C), and heart failure hospitalization or death (Panel D), according to tertiles of office systolic blood pressure variability.

When stratifying by BP control during the OBPV estimation period, higher OBPV examined as a continuous variable was associated with increased risk of CVD events, irrespective of BP control status. This persisted after covariate adjustment (BP controlled: HR=1.02, 95% CI: 1.00-1.04, P=0.048; BP not controlled: HR=1.02, 95% CI: 1.01-1.04 P=0.01; Table 3). However, upon dividing OBPV into tertiles, an increased risk of CVD was observed only in the highest tertile group in relation to the lowest tertile among those whose BP was not controlled (fully adjusted model HR=1.62, 95% CI: 1.20-2.20; P=0.002). In other analyses stratifying by use of antihypertensive drugs at baseline (Table S3), high OBPV was strongly associated with increased risk of CVD events, irrespective of whether or not participants were using antihypertensive drugs (receiving antihypertensives: T3, HR=1.38; 95% CI: 1.04-1.83; P=0.03; not receiving antihypertensives: T3, HR=1.56; 95% CI: 1.08-2.24; P=0.02). These findings were consistent across all models for OBPV as a continuous variable as well as in the highest OBPV tertile; for those participants who were not receiving antihypertensive drugs, increased risk of CVD was also observed in the middle tertile.

Table 3:

Association between office blood pressure variability (OBPV) and cardiovascular disease events among ASPirin in Reducing Events in the Elderly trial participants stratified by blood pressure control status during the OBPV estimation period (N=16,475).

	BP controlled (<140/90 mmHg)				BP not controlled (BP>140/90 mmHg)
	OBPV (continuous)	OBPV Tertiles			OBPV (continuous)	OBPV Tertiles
		T1	T2	T3		T1	T2	T3
N	9328	3146	3089	3093	7147	2413	2356	2379
Observed CVD events, N (%)	222 (2.38)	61 (1.94)	75 (2.43)	86 (2.78)	275 (3.85)	68 (2.82)	86 (3.65)	121 (5.09)
	Hazard Ratio (95% CI) P-value	Referent	Hazard Ratio (95% CI) P-value	Hazard Ratio (95% CI) P-value	Hazard Ratio (95% CI) P-value	Referent	Hazard Ratio (95% CI) P-value	Hazard Ratio (95% CI) P-value
*Model 1	1.04 (1.02-1.06) P<0.001	1.00	1.24 (0.88-1.74) P=0.22	1.38 (1.00-1.92) P=0.053	1.03 (1.01-1.05) P<0.001	1.00	1.27 (0.93-1.75) P=0.14	1.80 (1.34-2.43) P<0.001
Ɨ Model 2	1.03 (1.00-1.06) P=0.016	1.00	1.17 (0.84-1.65) P=0.35	1.20 (0.86-1.67) P=0.28	1.03 (1.01-1.04) P=0.004	1.00	1.28 (0.93-1.76) P=0.13	1.66 (1.23-2.25) P=0.001
ǂModel 3	1.02 (1.00-1.05) P=0.035	1.00	1.15 (0.82-1.62) P=0.41	1.15 (0.82-1.60) P=0.42	1.03 (1.01-1.04) P=0.005	1.00	1.28 (0.93-1.76) P=0.14	1.65 (1.22-2.24) P=0.001
§ Model 4	1.02 (1.00-1.04) P=0.048	1.00	1.15 (0.82-1.61) P=0.43	1.13 (0.81-1.58) P=0.48	1.02 (1.01-1.04) P=0.01	1.00	1.27 (0.92-1.75) P=0.14	1.62 (1.20-2.20) P=0.002

BP = blood pressure; OBPV = office blood pressure variability; T = tertile; CVD = cardiovascular disease; CI = confidence interval

^*Unadjusted

^ƗAdjusted for age, sex, diabetes, smoking, country, aspirin, ethnicity, body mass index, statin use at baseline

^ǂModel 2 plus adjusted for within-individual mean systolic blood pressure from baseline to second annual visit

^§Model 3 plus adjusted for use of antihypertensive drugs at baseline

Further sensitivity analysis using ARV to estimate OBPV revealed similar associations with CVD events, although the risk was no longer statistically significant when adjusted for within-individual mean systolic BP and use of antihypertensive drugs at baseline (Table S4). Lastly, sensitivity analysis estimating OBPV based on BP measurements from each of four visits (baseline, Years 1-3) revealed findings consistent with our primary results (fully adjusted model, T3 HR=1.37; 95% CI: 1.03-1.81; P=0.03) (Table S5 and Figure S2).

Discussion

In this post-hoc analysis of the ASPREE study, participants in the highest SD tertile for visit-to-visit systolic OBPV were at increased risk of the composite CVD endpoint, as well as ischemic stroke, hospitalization or death from heart failure, and all-cause mortality, relative to those individuals in the lowest tertile, even after adjustment for multiple covariates. Individuals in the highest tertile had a 36% increased risk of CVD during the follow-up period beyond the first two years of the study. Sensitivity analyses stratifying participants based on BP control status during the OBPV estimation period suggested that the increased risk was especially within individuals whose BP was uncontrolled during this period. Our results are important because they extend knowledge about the prognostic ability of OBPV to an older cohort consisting of individuals with and without hypertension who have reached their seventh, eighth and ninth decades of life without previous CVD events, and in an otherwise healthy and nondisabled state.

Although ASPREE was not a hypertension outcome study with a specified BP intervention, and included individuals without hypertension, our results are consistent with post-hoc analyses of large BP-lowering intervention studies conducted within high risk hypertensive populations. In the Anglo-Scandinavian Cardiac Outcomes Trial (ASCOT) Blood Pressure Lowering Arm, unadjusted analyses showed that participants in the highest decile of OBPV were 1.57 and 2.06 times more likely to suffer a coronary event or stroke compared to those in the lowest decile.⁴ In the Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT), a 1.30-fold increased risk of fatal or nonfatal coronary heart disease events was found in participants in the highest quartile of SD of OBPV.⁸ In contrast, OBPV was not associated with the composite end point of fatal and nonfatal cardiovascular events in the recent Systolic Blood Pressure Intervention Trial (SPRINT), although the highest quintile of OBPV was associated with increased all-cause mortality.⁹ Several reasons may explain the different SPRINT findings including the large contrast in BP between treatment groups which may have mitigated OBPV differences, while in ALLHAT there was very similar BP lowering across interventions and it was a much larger and longer study.

The mechanistic pathways underpinning the increased risk of CVD events in individuals with high OBPV remain elusive. Although there is usually some correlation between high OBPV and high BP, the increased risk of CVD events predicted by OBPV appears to be independent of mean BP.¹ Possible mechanisms may relate to morphologic alterations in the microcirculation including increased vascular stiffness in individuals with higher OBPV.¹⁵ A reciprocal relationship is possible though, as increased large artery stiffness associated with aging impairs baroreflex function, and increases cardiac workload, systolic BP, pulse pressure, and BP variability. A recent study examining coronary atheroma progression across 7 randomized clinical trials found that greater BP variability was associated with coronary atheroma progression and adverse clinical outcomes.¹⁶ The authors suggested that atheroma progression in the vasculature may be an inciting event leading to higher BP variability via underlying pathobiological mechanisms influenced by ischemia or impaired baroreflexes. Higher BP variability has been reported to increase oscillatory shear stress.¹⁷ Poor control of BP overall and changes in antihypertensive drugs can also increase BP variability.¹⁸ While there was some indication in our sensitivity analysis that the increased risk of CVD was primarily in those whose BP was not controlled during the OBPV estimation period, we observed poor correlation overall with average systolic BP which supports an independent predictive capability of OBPV.

Given the heterogeneity in phenotypes of older adults, an important research gap remains in ascertaining whether routinely targeting high OBPV with specific antihypertensive drug classes incrementally lowers CVD events beyond that expected when directing therapy toward average BP levels. A prospective study of such intent may be challenging to design in order to clearly differentiate the effects of reducing OBPV apart from lowering of mean BP; until then, post-hoc analyses of existing cohorts can only be speculative. While OBPV typically decreases along with average BP in response to treatment, it has been recommended to avoid short-acting antihypertensive drugs due to their propensity to induce iatrogenic increases in BP variability.¹ There is some evidence that certain drug classes, namely long-acting calcium channel blockers, such as amlodipine, and the thiazide-like diuretic, chlorthalidone, are associated with reduced long-term BP variability relative to other classes.^19-22 Agents such as these may be preferred for reducing OBPV due to their ability to lower BP consistently throughout the twenty-four hour dosing period. Ensuring optimal patient adherence is also important to minimizing variability in BP.²³

The large sample size of ASPREE, use of standardized BP measurements, adjudication of all CVD events, and our analytical approach to diminish the impact of immortal time bias, are key strengths of our study. Most importantly, most other studies of visit-to-visit OBPV have been conducted within cohorts from hypertension intervention studies, while ASPREE participants were generally healthy at baseline and therefore more likely to reflect the real-world risk of high OBPV in older aged, community-dwelling adults.

These strengths must be balanced with the important limitations of our analysis. These include our inability to account for medication non-adherence or changes in antihypertensive regimens, which might increase or influence BPV.²³ However, of those individuals using antihypertensive medications at baseline, >99% continued use during the trial (data not shown). Although we used only three BP records to estimate OBPV, there is no accepted consensus on the minimum number of BP records which should be used to estimate OBPV, and our results were consistent in sensitivity analysis which used an alternate criteria of four BP records. Despite controlling for covariates associated with OBPV and CVD, the potential impact of unmeasured confounders must be acknowledged. Finally, our findings cannot establish causality, and whether lowering OBPV impacts CVD events remains an important unresolved question.

Perspectives

In this post-hoc analysis of the ASPREE trial, high OBPV estimated using BP measures taken from baseline through the second annual visit was associated with increased risk of the CVD composite during follow-up, as well as ischemic stroke, hospitalization or death from heart failure, and all-cause mortality. These findings persisted after controlling for a large number of covariates, including mean BP, and across several sensitivity analyses. While directing therapy towards lowering mean BP remains the most proven strategy to reduce CVD risk, our findings strengthen support for consideration of long-term OBPV as a potential clinically actionable parameter in older adults with and without hypertension.

Novelty and Significance.

What is New?

• We explored the association of long-term, visit-to-visit office blood pressure (BP) variability (OBPV) with cardiovascular disease (CVD) events in a cohort of community-dwelling adults who averaged 75 years of age.

• Since it was not a BP intervention study, investigating OBPV in the ASPirin in Reducing Events in the Elderly (ASPREE) cohort potentially increases the generalizability of the findings to a general population of older adults.

What is Relevant?

• ASPREE included individuals with, and without hypertension, who had reached older age, in otherwise good health, and without dementia, physical disability, or prior CVD events.

• Higher OBPV was associated with increased risk of CVD events, as well as ischemic stroke, heart failure hospitalization or death, and all-cause mortality, even after controlling for key covariates.

Summary

• Older adults without previous CVD who have high OBPV are at increased risk for CVD events in their remaining lifetimes. Future research should examine whether lowering OBPV reduces CVD event rates beyond that expected when lowering mean BP.