Office, central and ambulatory blood pressure for predicting first stroke in older adults: A community-based cohort study

Abstract

Hypertension is the most prevalent modifiable risk factor for stroke. Office blood pressure (BP) measurements may have limitations in defining the impact of hypertension on stroke. Our aim was to compare the stroke risk for office, central, and ambulatory BP measurements in a predominantly older population-based prospective cohort. Participants in the Cardiovascular Abnormalities and Brain Lesions (CABL) study (n=816; mean age, 70.8±9.0 years; 39.8% men) underwent applanation tonometry of the radial artery for central BP and 24-hour ambulatory BP monitoring. During a follow-up of 9.6±3.1 years, stroke occurred in 76 participants (9.3%). Among office BP variables, only diastolic BP was associated with stroke in multivariable competing risk model (P=0.016). None of the central BP variables showed a significant association with stroke. Conversely, all ambulatory systolic and diastolic BP variables were significantly associated with stroke after adjustment for clinical confounders (all P<0.005). In an additional multivariable competing risk model including both ambulatory systolic and diastolic BP values obtained at the same time of the day, diastolic BP was more strongly associated with stroke than systolic BP in 24-h, daytime and nighttime periods (all P<0.05). Therefore, in a predominantly older population-based cohort, office diastolic BP was weakly associated with incident stroke; no central BP variable was prognostic of stroke. However, all ambulatory systolic and diastolic BP values were significantly associated with stroke in multivariable competing risk analyses. Moreover, ambulatory diastolic BP was a stronger predictor of stroke than ambulatory systolic BP.

Introduction

Stroke is a leading cause of long-term disability and the second-leading cause of mortality worldwide.^1,2 Over the next several decades, the number of strokes is expected to more than double, with most of the increase occurring among older adults.¹ Older stroke patients experience longer hospitalization, greater disability and higher mortality²; therefore, primary stroke prevention is critically important for older adults.

Hypertension is the most prevalent modifiable risk factor for stroke. Although high blood pressure (BP) is known to have significant effects on all cardiovascular outcomes, lowering BP reduced stroke incidence to a greater extent than coronary artery disease and mortality in a recent meta-analysis.³ Previous post-hoc analyses, usually from studies on the efficacy of antihypertensive drugs, indicated a significant association between high office BP and incident stroke.^4–6 However, conflicting results have been reported in recent randomized controlled trials regarding the impact of achieving BP targets, based on office BP, for the prevention of stroke.^4,7

More sophisticated BP measurements, such as central BP⁸ and ambulatory BP,⁹ have proven to be stronger predictors of future cardiovascular events than conventional office BP. Several studies indicated that BP values obtained at the level of aorta (central BP) had a stronger correlation with stroke than peripheral BP did.¹⁰ In addition, higher ambulatory BP was associated with an increased risk for stroke independently of office BP in community-based studies.^11–13 However, those studies were limited by the relatively young age of participants. Furthermore, few studies explored both central and ambulatory BP variables as potential predictors of first stroke in the same individuals. Accordingly, the objective of the present study was to compare the prognostic impact for stroke of office, central, and ambulatory BP measurements in a predominantly older multiethnic population-based cohort without history of stroke.

Methods

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Study population

The study population was derived from the Cardiovascular Abnormalities and Brain Lesions (CABL) study, which was designed to investigate the cardiovascular predictors of subclinical cerebrovascular disease in a community-based cohort including participants over age 55. The cohort of the CABL study based its recruitment on the Northern Manhattan Study (NOMAS), an observational, prospective, population-based cohort of stroke-free participants that enrolled from the Northern Manhattan neighborhood between 1993 and 2001. The study design and methodological details regarding NOMAS have been described previously.¹⁴ Of a total of 1,004 participants enrolled in CABL, 995 successfully underwent applanation tonometry of the radial artery, and 835 had ambulatory BP monitoring performed; 816 had both assessments available and represent the cohort of the present study. The investigation conforms with the Declaration of Helsinki. Written informed consent was obtained from all study participants. The study was approved by the Institutional Review Boards of Columbia University Medical Center and the University of Miami.

Risk factor assessment and office BP measurement

Detailed information on the cardiovascular risk factors in the cohort is shown in the online-only Data Supplement. Both office systolic BP (SBP) and diastolic BP (DBP) were measured on the non-dominant arm in sitting position after five minutes of rest, using a mercury sphygmomanometer and an arm cuff of appropriate size. Office BPs were recorded twice with a 5-minute interval, and the average of the two recordings was used. Office pulse pressure (PP) was defined as the difference between SBP and DBP.

Central BP measurement

For determination of central BPs, pulse wave analysis of the radial artery by applanation tonometry tracing was performed using a commercially available device (SphygmoCor, Pulse Wave Analysis System, AtCor Medical, Sydney, Australia). A detailed description of the technique and reproducibility data have been previously published.¹⁵ Central SBP, DBP, and PP were calculated from the radial pulse wave by a validated generalized transfer function.¹⁵

Ambulatory BP measurement

An ambulatory BP monitor (SpaceLabs Model 90207, Snoqualmie, WA) was used to assess 24-h BP as the subjects performed their normal activities. Ambulatory BP monitoring was performed according to previously published protocols with a BP cuff appropriately sized to arm circumference and placed on the non-dominant arm.¹⁵ A BP reading was automatically taken and recorded every 15 minutes during awake hours and 30 minutes during sleep hours. Recordings were retrieved and analyzed with the aid of ambulatory BP report management system software (SpaceLabs Systems, 2004). The average SBP, DBP and PP were calculated for a 24-h period and separately for daytime (awake) and nighttime (sleep) periods, which were determined using subjects’ diary reports of sleep and wake times.

Follow-up and outcome evaluation

All subjects were followed annually by telephone. Any vascular event or acknowledgment of neurological or cardiac symptoms during the annual standardized interview triggered an in-person assessment. All subjects who screened positive for possible stroke were assessed in person by a neurologist. In addition, active hospital surveillance of admission and discharge International Classification of Diseases-9th revision codes was performed. Stroke was defined by the first symptomatic occurrence of any type of stroke including infarcts and hemorrhages. Ischemic strokes were defined by TOAST (Trial of ORG 10172 in Acute Stroke Treatment) criteria.¹⁶ Diagnosis of stroke was determined by consensus of two neurologists. Both fatal and nonfatal stroke were used as the primary outcome of the present study. Time to first stroke was calculated from the study enrollment to stroke.

Statistical analysis

Continuous data are presented as mean±standard deviation and categorical variables as frequencies and percentages. Fine and Gray’s proportional subdistribution hazards models¹⁷ were used to assess the association of BP parameters with stroke, and hazard ratios and 95% confidence intervals were calculated. The occurrence of death from any causes other than stroke was treated as a competing event. Variables associated with stroke in univariable analyses with a probability value 0.10 or less were entered as covariates in multivariable models. All calculations were performed using SAS version 9.3 (SAS Institute, Cary, NC) and P values less than 0.05 were considered statistically significant.

Results

Study population

The baseline characteristics of the study population are shown in Table 1. The mean age of the study population was 70.8±9.0 years, 39.8% of participants were men, 69.9% were Hispanic, 15.8% were black, 12.5% were white, and 1.8% of other race/ethnicities. Overall, 638 participants (78.2%) had hypertension at baseline. The baseline characteristics of participants included or excluded or from the analysis was shown in Table S1 of the online-only Data Supplement.

Table 1.

Baseline characteristics of the study cohort

Characteristic	n=816
Age, years	70.8±9.0
Male sex	325 (39.8%)
Race
White	102 (12.5%)
Black	129 (15.8%)
Hispanic	570 (69.9%)
Others	15 (1.8%)
Hypertension	638 (78.2%)
Number of antihypertensive drugs (n=630)
0	192 (30.5%)
1	208 (33.0%)
2	164 (26.0%)
3	57 (9.1%)
4	9 (1.4%)
Antihypertensive drug class
ACEi/ ARB (n=659)	238 (36.1%)
Beta-blocker (n=816)	206 (25.3%)
Calcium channel blocker (n=650)	230 (35.4%)
Diuretics (n=648)	167 (25.8%)
Hypercholesterolemia	559 (68.5%)
Diabetes mellitus	240 (29.4%)
Body mass index, kg/m2	28.3±4.8
Coronary artery disease	50 (6.1%)
Cigarette smoking	430 (52.7%)
eGFR, mL/min/1.73m2 (n=790)	73.9±18.3
Atrial fibrillation	50 (6.1%)
Education
8th grade or less	331 (40.6%)
some high school	122 (15.0%)
completed high school	115 (14.1%)
some college	117 (14.3%)
college graduate or more	131 (16.1%)
Office BP variables
Office SBP, mm Hg	135.6±17.7
Office DBP, mm Hg,	78.6±9.4
Office PP, mm Hg	57.0±14.8
Central BP variables
Central SBP, mm Hg	119.5±18.9
Central DBP, mm Hg	72.1±10.2
Central PP, mm Hg	47.4±15.3
Ambulatory BP variables
24-h SBP, mm Hg	124.8±14.4
24-h DBP, mm Hg	71.4±8.6
24-h PP, mm Hg	53.5±11.2
Daytime SBP, mm Hg	128.2±14.5
Daytime DBP, mm Hg	74.2±9.0
Daytime PP, mm Hg	54.0 ±11.3
Nighttime SBP, mm Hg	118.7±16.3
Nighttime DBP, mm Hg	66.1±9.4
Nighttime PP, mm Hg	52.6±11.9

Values are mean±standard deviation or n (percentage). ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; BP, blood pressure; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; PP, pulse pressure; SBP. systolic blood pressure.

Clinical variables and stroke

Mean follow-up was 9.6±3.1 years. Stroke occurred in 76 participants (9.3%). During follow-up, 224 (27.5%) participants died without having experienced a stroke; therefore, death from any causes other than stroke was treated as a competing risk in all subsequent survival analyses. The associations of clinical variables with stroke using univariate Cox proportional hazards regression analyses are shown in Table 2. Age, never-smoker status, presence of hypertension, hypercholesterolemia, chronic kidney disease and atrial fibrillation were significantly associated with stroke (all P<0.05). The use of two or more antihypertensive drugs showed a non-significant trend towards an association with stroke (P=0.066).

Table 2.

Association of clinical variables with stroke in univariable proportional subdistribution hazards regression analysis

Variable	SHR (95%CI)	P
Age, year	1.06 (1.03―1.09)	<0.001
Sex, male	0.94 (0.59―1.50)	0.80
Race
White	1 (reference)
Black	0.82 (0.39―1.74)	0.61
Hispanic	0.64 (0.35―1.19)	0.16
Hypertension	2.17 (1.08―4.36)	0.029
Two or more antihypertensive drugs use	1.55 (0.97―2.46)	0.066
Hypercholesterolemia	2.60 (1.41―4.77)	0.002
Diabetes mellitus	1.20 (0.75―1.94)	0.45
Body mass index, kg/m2	0.99 (0.95―1.04)	0.78
Coronary artery disease	1.60 (0.74―3.46)	0.23
Cigarette smoking	0.60 (0.38―0.95)	0.030
Chronic kidney disease	2.04 (1.26―3.30)	0.004
Atrial fibrillation	5.13 (2.99―8.79)	<0.001
Some college or more	1.01 (0.62—1.64)	0.98

CI, confidence interval; SHR, subdistribution hazard ratio.

BP values and stroke

Univariable and multivariable proportional subdistribution hazards regression analyses were performed to identify BP variables associated with stroke (Table 3). For office BP variables, DBP was the only variable associated with stroke in multivariable models. None of the central BP variables showed a significant association with stroke in either univariable or multivariable analyses. All ambulatory BP variables were significantly associated with stroke in an unadjusted model (all P<0.05). After adjusting for several clinical confounders (model 1), all ambulatory SBP and DBP variables remained associated with stroke (all P<0.05). These associations persisted after further adjustment for use of two or more antihypertensive drugs (multivariable model 2; all P<0.05). None of the ambulatory PP variables showed a significant association with stroke. Also, none of the additional parameters obtained from ambulatory BP monitoring (circadian variability; night-to-day ratio; morning surge) was associated with stroke in the fully adjusted models (Table S2). Furthermore, additional analyses failed to show a clear non-linear relationship (U-shaped or J-shaped) with stroke of any BP variables in our cohort (Figure S1–3).

Table 3.

Association of office, central and ambulatory blood pressure variables with stroke – univariable and multivariable proportional subdistribution hazards regression analyses

BP variables	Univariable model		Multivariable model 1†		Multivariable model 2‡
	SHR (95%CI)*	P	SHR (95%CI)*	P	SHR (95%CI)*	P
Office BP variables
Office SBP	1.15 (1.02―1.31)	0.027	1.12 (0.97―1.29)	0.11	1.12 (0.97―1.30)	0.13
Office DBP	1.23 (0.98―1.54)	0.074	1.32 (1.05―1.66)	0.016	1.33 (1.04―1.69)	0.022
Office PP	1.13 (0.96―1.34)	0.13	1.05 (0.88―1.25)	0.60	1.05 (0.87―1.25)	0.63
Central BP variables
Central SBP	1.01 (0.90―1.14)	0.82	1.01 (0.89―1.14)	0.91	1.00 (0.89―1.13)	0.94
Central DBP	1.00 (0.81―1.25)	0.99	1.22 (0.95―1.56)	0.12	1.22 (0.95―1.56)	0.13
Central PP	1.02 (0.88―1.18)	0.79	0.93 (0.78―1.09)	0.36	0.92 (0.78―1.09)	0.34
Ambulatory BP variables
24-h SBP	1.28 (1.13―1.46)	<0.001	1.28 (1.11―1.47)	<0.001	1.29 (1.10―1.51)	0.002
24-h DBP	1.37 (1.08―1.73)	0.010	1.65 (1.30―2.11)	<0.001	1.68 (1.29―2.19)	<0.001
24-h PP	1.27 (1.07―1.51)	0.006	1.12 (0.92―1.37)	0.26	1.12 (0.91―1.38)	0.28
Daytime SBP	1.24 (1.09―1.42)	<0.001	1.24 (1.08―1.43)	0.002	1.25 (1.08―1.45)	0.003
Daytime DBP	1.29 (1.03―1.62)	0.026	1.59 (1.26―2.01)	<0.001	1.61 (1.25―2.06)	<0.001
Daytime PP	1.23 (1.04―1.46)	0.019	1.08 (0.89―1.32)	0.44	1.08 (0.88―1.32)	0.46
Nighttime SBP	1.25 (1.12―1.40)	<0.001	1.25 (1.10―1.42)	<0.001	1.26 (1.09―1.46)	0.002
Nighttime DBP	1.38 (1.11―1.72)	0.004	1.53 (1.23―1.90)	<0.001	1.57 (1.23―2.00)	<0.001
Nighttime PP	1.26 (1.08―1.48)	0.004	1.14 (0.95―1.38)	0.17	1.14 (0.94―1.38)	0.19

BP, blood pressure; CI, confidence interval; DBP, diastolic blood pressure; PP, pulse pressure; SBP, systolic blood pressure; SHR, subdistribution hazard ratio.

^*Subdistribution hazard ratios express the risk for increments of 10 mm Hg.

^†Adjusted for age, race, hypercholesterolemia, cigarette smoking, chronic kidney disease and atrial fibrillation.

^‡Adjusted for age, race, hypercholesterolemia, cigarette smoking, chronic kidney disease, atrial fibrillation and two or more antihypertensive drugs use.

To assess whether ambulatory SBP or DBP was more important for stroke prediction, we simultaneously entered SBP and DBP values obtained at the same time of the day into an additional multivariable model (Table 4). DBP was more strongly associated with stroke than SBP in 24-h, daytime and nighttime periods (all P<0.05). Comparisons of the models with likelihood ratio test to assess the predictive value of ambulatory BP for incident stroke are shown in Table S3.

Table 4.

Comparison of the association of ambulatory systolic and diastolic blood pressure with stroke

SBP vs. DBP	Multivariable model 1†		Multivariable model 2‡
	SHR (95%CI)*	P	SHR (95%CI)*	P
24-h SBP vs. 24-h DBP
24-h SBP	1.06 (0.86―1.32)	0.58	1.07 (0.86―1.33)	0.54
24-h DBP	1.54 (1.07―2.21)	0.22	1.55 (1.07―2.25)	0.021
Daytime SBP vs. daytime DBP
Daytime SBP	1.03 (0.83―1.28)	0.79	1.03 (0.83―1.28)	0.77
Daytime DBP	1.54 (1.07―2.21)	0.020	1.55 (1.07―2.24)	0.020
Nighttime SBP vs. nighttime DBP
Nighttime SBP	1.07 (0.88―1.30)	0.48	1.08 (0.89―1.32)	0.45
Nighttime DBP	1.41 (1.02―1.94)	0.038	1.43 (1.02―2.00)	0.036

BP, blood pressure; CI, confidence interval; DBP, diastolic blood pressure; SBP, systolic blood pressure; SHR, subdistribution hazard ratio.

^*Subdistribution hazard ratios express the risk for increments of 10 mm Hg.

^†Systolic and diastolic blood pressure are included in the same model adjusted for age, race, hypercholesterolemia, cigarette smoking, chronic kidney disease and atrial fibrillation.

^‡Systolic and diastolic blood pressure are included in the same model adjusted for age, race, hypercholesterolemia, cigarette smoking, chronic kidney disease, atrial fibrillation and two or more antihypertensive drugs use.

Because 438 participants (53.7%) used at least one antihypertensive drug, which might affect the association between BP variables and stroke, we performed a sensitivity analysis limited to the 378 participants who were not receiving antihypertensive medications (Table S4). In this analysis, office DBP and all ambulatory DBP variables remained independently associated with stroke in multivariable models (all P<0.05).

Because 188 of the 1,004 CABL participants had no ambulatory BP information and were therefore excluded from the main analysis, and since they differed from those included with regard to some pertinent characteristics (Table S1), we performed a sensitivity analysis on the risk of stroke associated with office BP variables in the entire cohort; this analysis showed that none of the office BP values was associated with stroke in multivariable models (Table S5).

Finally, the prognostic impact of type of hypertension is shown in Table S6. The relationship of masked hypertension with stroke was stronger than that of normotension, but did not reach statistical significance in the multivariable analysis (P=0.085). Only sustained hypertension was associated with stroke in the fully adjustment models (P<0.001).

BP values and death

Univariable Cox proportional hazards regression analysis showed that age, male sex, non-Hispanic race/ethnicity, the use of two or more antihypertensive drugs, body mass index, presence of hypertension, diabetes mellitus, coronary artery disease, chronic kidney disease and atrial fibrillation were significantly associated with death (all P<0.05; Table S7). Although neither office nor central BP variables showed a significant association with death after adjustment for clinical covariates, ambulatory BP variables, including 24-h and nighttime SBP and PP, were associated with death (Table S8).

Discussion

Office BP and stroke

Hypertension is the most important modifiable risk factor for stroke. High BP leads to extensive alteration in endothelium and smooth muscle function in the intracerebral arteries, which can cause both ischemic and hemorrhagic lesions. In addition, hypertension accelerates the arteriosclerotic process, thus increasing the likelihood of cerebral lesions related to stenosis and embolism originating from large extracranial vessels, aortic arch and from the heart.¹⁸ These mechanisms may explain the results of meta-analyses of randomized clinical studies comparing hypertension treatments with different BP targets, which have suggested the efficacy of more intensive control of office BP.^5,6 However, it should be noted that two clinical trials reached opposite conclusions.^4,7 The Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial showed the association of an intensive BP control (SBP target of <120 mm Hg compared with the standard SBP target of <140 mm Hg) with a reduction in the risk of stroke.⁴ On the other hand, there was no difference in stroke occurrence between the two BP control groups in SPRINT (Systolic Blood Pressure Intervention Trial).⁷ Despite the similar study design of the two trials, the older age of the study group and the exclusion of prior stroke in the SPRINT may have affected the results. Among CABL participants, we did not observe a significant association between office SBP and incident stroke, but only a weak relationship of office DBP with stroke. As with the SPRINT trial, the definition of stroke outcome and the different demographic composition, including the older age of our cohort, might account for the lack of association observed. Also, the exclusion of diabetes may have been a factor in weakening the association between office SBP and stroke observed in the SPRINT trial, because high BP and diabetes interacts synergistically and greatly increase the stroke risk.¹⁹ Therefore, this may also help explain the lack of a significant relationship between office SBP and stroke in CABL, a community-based cohort including almost 30% of participants with diabetes. Moreover, since patients with prior stroke were excluded from CABL, lower than expected stroke rates for age may have occurred, which may have weakened our ability to detect an association between office SBP and stroke. However, the exclusion of individuals with prior stroke provided us with the possibility to assess the relationship of BP with first-ever stroke, not confounded by the presence a strong risk factor for a recurrent stroke.

In our study, PP was not associated with incident stroke. Aging-related aortic stiffening leads to increase in SBP, from both stiffening of the proximal aorta and early return of the reflected wave, and to a decrease in DBP.²⁰ As a result, the increased PP reflects large artery stiffness, whereas aging also contributes to the increase of PP.²¹ Accordingly, because of the increased aortic stiffness secondary to aging, PP may be a less effective risk predictor in older adults; therefore, this circumstance may be involved in explaining the lack of a significant association between PP and stroke that we observed.

Central BP and stroke

Several studies have shown central BP to be a better prognostic indicator of composite cardiovascular outcomes compared with brachial BP measured in the doctor’s office.⁸ However, the association between central BP and stroke is not fully elucidated. Recently, Chuang et al showed that central SBP independently predicted incident stroke in a community-based cohort without stroke history.¹⁰ In contrast, we did not observe a significant relationship of central BP with incident stroke. Possibly because aging promotes near maximal stiffening of the large arteries independent of atherosclerotic disease severity, central BPs did not show a significant association with stroke in older adults. This finding is consistent with a previous study.²² Furthermore, the higher frequency in our cohort of antihypertensive medication use, such as angiotensin-converting enzyme inhibitors, angiotensin receptor blockers and calcium channel blockers, which have the potential to lower central BP,⁸ may have affected our ability to detect an independent role of central BP on stroke risk. However, the same result was also observed in the sensitivity analysis only including untreated participants (Table S4), suggesting an actual lack of independent effect of central BP on stroke risk.

Ambulatory BP and stroke

Ambulatory BP monitoring, which allows the recording of BP throughout the day and night during an individual’s usual activities, is considered the gold standard for assessing BP status.²³ In our cohort, the association of ambulatory BP with incident stroke was the strongest among all considered BP parameters. These findings are consistent with previous studies that reported on how ambulatory BP values were associated with cardiovascular events including stroke.^11–13 However, stroke was assessed as a part of a composite primary outcome and not as a separate endpoint in most of those studies;^11,13 few analyses have been published evaluating the prognostic impact of ambulatory BP on first stroke.^11,12 Fagard et al demonstrated that any of ambulatory SBP and DBP variables independently predicted future stroke among hypertensive patients without a history of cardiovascular disease from four studies performed in Europe.¹² Similarly, ambulatory BP was more strongly associated with first stroke than office BP in a Portuguese hypertensive cohort.¹¹ However, these studies were limited by the relatively young age of participants.^11,12 Our results confirm the impact of ambulatory BP variables on stroke in predominantly older adults, in whom primary stroke prevention is critically important. Moreover, to the best of our knowledge the present is the first study in which office, central and ambulatory BP were explored as potential first stroke predictors among the same individuals of a population-based cohort.

DBP and stroke

Hypertension is specifically implicated in stroke subtypes related to small vessel disease, including lacunar infarction and intracerebral hemorrhage. In fact, cerebral small vessels are particularly exposed to sustained BP elevations easily transmitted from the aorta.²⁰ Accordingly, a high DBP, indicating elevated steady blood flow, may lead to remodeling in small arteries and arterioles, which may predispose to luminal abnormalities and ischemia.²⁴ This possible underlying mechanism may explain why DBP was a stronger stroke predictor than SBP in the present study. Also, our finding is consistent with a previous report from NOMAS, in which DBP, but not SBP, was associated with subclinical cerebrovascular disease detected by neuroimaging.²⁴ Furthermore, both ambulatory DBP and SBP were associated with an increased risk for future stroke in population-based studies.^11–13 Nevertheless, few studies to date have addressed the relative importance of ambulatory SBP and DBP in predicting cardiovascular outcomes. In the present study, the prognostic significance of ambulatory DBP values persisted after adjustment for corresponding SBP levels. On the contrary, Mesquita-Bastos et al showed that the predictive value of ambulatory SBPs for stroke persisted significantly after further adjustment for the correspondent values of ambulatory DBP.¹¹ In addition to the possible mechanism for DBP and stroke mentioned above, the lower mean ambulatory SBP observed in our cohort (by approximately 10 mm Hg compared with the Portuguese cohort) might have weakened our ability to detect an independent role of ambulatory SBP for predicting stroke when DBP was also considered. Indeed, it is noteworthy that Chen et al recently showed a significant association between DBP and first stroke even after adjustment for SBP in a Chinese community-based cohort that showed relatively low BP values, at least for older hypertensive individuals.²⁵ Further studies are needed to confirm whether these findings may be generalized to cohorts with different demographics.

Strengths and limitations

The main strengths of the present study are its prospective design, the long follow-up duration, the rigorous stroke adjudication by neurologists, the relatively large population-based sample of participants undergoing three different techniques of BP measurement, and the extensive adjustment for possible confounders. However, our study also has limitations. First, although representative of the multi-ethnic community living in Northern Manhattan, the study participants were predominantly older and disproportionately of Hispanic ethnicity. Therefore, the study findings might not apply to populations with different demographic profiles. Also, participants for CABL were originally selected to participate in a neuroimaging study of the brain, which may have selected more health-conscious individuals. Furthermore, the exclusion of individuals with a prior stroke, in whom the association between BP and stroke may be different, may introduce a selection bias that affects the inferences from the results. Second, the frequency of hypertension and antihypertensive drug use in our cohort was relatively high, as expected in a cohort of older adults. Although a significant association between ambulatory BP variables and stroke persisted even after adjustment for antihypertensive treatment and was also observed in untreated participants, our results might not be directly applicable to cohorts with different risk factors profiles or higher proportions of normotensive participants. Third, follow-up BP measurements were not uniformly available. However, our aim was to evaluate which one-time assessment of BP status would better predict stroke risk, rather than how changes in BP or effect of treatment would impact on the prediction. Finally, according to the method for measuring office BP in the current ESC/ESH guidelines, three or more readings 1–2 minutes apart are recommended, and then BP is recorded as the average of the last two readings.²⁶ Therefore, the modalities of BP measurement in our study (average of 2 measurements 5 minutes apart) may have affected the results obtained and possibly the stroke predictive power of office BP.

Perspectives

In a predominantly older multiethnic population-based cohort without history of stroke, office DBP was weakly associated with incident stroke; no central BP variable was predictive of stroke. However, all ambulatory BP values were significantly associated with stroke risk in multivariable analyses. Furthermore, ambulatory DBP was independently associated with incident stroke even after adjustment for ambulatory SBP, indicating its stronger predictive value. Based on our results, the widespread use of ambulatory BP monitoring in older adults may be recommended to replace office BP for improving risk stratification for future stroke. While ambulatory BP monitoring is indicated for the diagnosis of hypertension in the 2017 ACC/AHA hypertension guidelines,²³ a recent meta-analysis showed that the clinical meaningful risk associated with elevated ambulatory BP in patients with controlled office BP extends to those receiving antihypertensive treatment.²⁷ Given the relatively high frequency of hypertension and antihypertensive drug use in CABL, our results also support the relevance of ambulatory BP in the prognostic stratification of treated patients. Moreover, controversies still exist regarding the use of a DBP threshold for defining hypertension and initiating treatment, due to the possible existence of a J-shaped relationship between DBP and adverse cardiovascular outcomes.²⁸ Although high-risk patients have a treatment threshold set at 130/80 mm Hg in the current hypertension guidelines, no DBP treatment threshold or target was recommended for older adults.²³ Given the results of the present study, future research evaluating the additional impact of ambulatory DBP reduction on the occurrence of stroke might be helpful in planning preventive strategies.

Novelty and Significance.

1). What is New?

• Few studies explored both central and ambulatory BP variables as potential predictors of first stroke in the same individuals.

• This study sought to investigate the stroke risk for office, central, and ambulatory blood pressure measurements in a predominantly older population-based prospective cohort.

2). What is Relevant?

Our findings support the hypothesis that ambulatory BPs are more directly involved among office, central and ambulatory BPs in the pathogenesis of cardiovascular disease.

3). Summary

In a predominantly older population-based cohort, office DBP was weakly associated with incident stroke; no central BP variable was prognostic of stroke. However, all ambulatory BP values were significantly associated with stroke risk in multivariable analyses.