Linear prediction plots and 95% CIs for the association between systolic BP variability (SD) and echocardiographic outcomes: left-ventricular (LV) mass, global longitudinal strain, and higher mitral inflow velocity to early diastolic mitral annular velocity (E/é).
eTable 1. Characteristics of CARDIA Participants at the Year-0 (1985-1986) by Inclusion in Blood Pressure Variability Analysis
eTable 2. Association Between Visit-to-Visit Systolic Blood Pressure Variability and Cardiac Structure and Function Among Participants Who Never Used Anti-Hypertensive Medication and Had Average Blood Pressure <130/80 mm Hg
eTable 3. Association Between Visit-to-Visit Diastolic Blood Pressure Variability and Cardiac Structure and Function Among Participants Who Never Used Anti-Hypertensive Medication and Had Average Blood Pressure <130/80 mm Hg
eTable 4. Sensitivity Analysis Showing the Association Between Systolic Diastolic Blood Pressure Variability and Cardiac Target Organ Damage Phenotypes (Left Ventricular Hypertrophy, Diastolic Dysfunction, and Left Atrial Enlargement)
eFigure. Histogram Showing the Distribution of Systolic and Diastolic Blood Pressure Variability in Analytic Sample
Customize your JAMA Network experience by selecting one or more topics from the list below.
Nwabuo CC, Yano Y, Moreira HT, et al. Association Between Visit-to-Visit Blood Pressure Variability in Early Adulthood and Myocardial Structure and Function in Later Life. JAMA Cardiol. 2020;5(7):795–801. doi:10.1001/jamacardio.2020.0799
Is there an association between visit-to-visit blood pressure variability in early adulthood and myocardial structure and function in middle age?
In this cohort study including 2400 participants, greater visit-to-visit variability in systolic and diastolic blood pressure in young adulthood was associated with higher left-ventricular mass, higher left-ventricular filling pressures, impaired left-ventricular relaxation, and greater global longitudinal strain.
Greater visit-to-visit systolic and diastolic blood pressure variability in early adulthood appears to be associated with adverse alterations in cardiac structure as well as systolic and diastolic function in middle age.
Long-term blood pressure (BP) variability has emerged as a reproducible measure that is associated with heart failure independent of systemic BP. Visit-to-visit BP variability may be associated with the risk of heart failure early in the life course and thus may be reflected in subclinical alterations in cardiac structure and function.
To evaluate the association between visit-to-visit BP variability in early adulthood and myocardial structure and function in middle age.
Design, Setting, and Participants
This cohort study used data from the Coronary Artery Risk Development in Young Adults (CARDIA) study, a community-based cohort study of 5115 participants aged 18 to 30 years at baseline (year 0; March 25, 1985, to June 7, 1986) and followed up over a 30-year interval. A total of 2400 CARDIA study participants underwent evaluation at 4 field sites (Birmingham, Alabama; Oakland, California; Chicago, Illinois; and Minneapolis, Minnesota). Blood pressure was measured at 8 visits over a 25-year interval and participants received echocardiograms at year 25 (June 1, 2010, to August 31, 2011). Data analysis was conducted from June 7, 1986, to August 31, 2011.
Visit-to-visit systolic and diastolic BP variability measures included SD, average real variability, and variability independent of the mean.
Main Outcomes and Measures
Echocardiographic indices of myocardial structure, systolic function, and diastolic function at the year 25 examination.
Of the 2400 participants, 1024 were men (42.7%) and 976 were African American (40.7%); mean (SD) age at the year 25 examination was 50.4 (3.6) years. Per 1-SD increment, greater visit-to-visit systolic BP variability independent of the mean was associated with higher left-ventricular (LV) mass index (β [SE], 2.66 [0.4] g/m2, P < .001), worse diastolic function (early peak diastolic mitral annular velocity [é]) (β [SE], −0.40 [0.1] cm/s, P < .001), higher LV filling pressures (mitral inflow velocity to early diastolic mitral annular velocity [E/é]) β [SE], 0.37 [0.1] cm/s, P < .001), and worse global longitudinal strain (β [SE], 0.17 [0.1], P = .002). Similarly, greater visit-to-visit diastolic BP variability was associated with higher LV mass index (β [SE], 3.21 [0.5] g/m2, P < .001), worse diastolic function (é: β [SE], −0.24 [0.1] cm/s [P < .001]; E/é: β [SE], 0.23 [0.1] cm/s [P < .001]), and worse global longitudinal strain (β [SE], 0.13 [0.1], P = .02). The findings remained consistent when other BP variability measures were used (SD and average real variability).
Conclusions and Relevance
In this cohort study using data from the CARDIA study, greater visit-to-visit systolic and diastolic BP variability have been associated with adverse alterations in cardiac structure as well as systolic and diastolic function independent of mean BP levels.
Blood pressure (BP) is an inherently dynamic physiologic measure that can fluctuate over time. Rather than representing measurement error in the assessment of an individual’s true BP, long-term visit-to-visit BP variability has emerged as a risk measure that can be reliably reproduced.1
Evidence suggests that, independent of systemic BP levels, greater visit-to-visit BP variability is associated with an increased risk of heart failure.2,3 The underlying mechanisms that support observed associations between long-term BP variability and heart failure are, however, incompletely understood. In this context, it is established that cardiac remodeling precedes the onset of cardiovascular disease, such as heart failure.4 Potential associations between visit-to-visit BP variability and alterations in myocardial structure and function may therefore further elucidate reported associations with cardiovascular disease outcomes, particularly heart failure. We therefore investigated the association between visit-to-visit systolic and diastolic BP variability and echocardiographic features associated with heart failure.
This cohort study used data from the Coronary Artery Risk Development in Young Adults (CARDIA) study; the design and participant selection of that study has been previously published.5 In summary, the CARDIA study is a community-based cohort study that recruited 5115 white and African American men and women participants from 4 US field sites (Birmingham, Alabama; Oakland, California; Chicago, Illinois; and Minneapolis, Minnesota). The study was approved by the institutional review boards at each site, and all participants provided written informed consent. Participants did not receive financial compensation.
Participants were aged 18 to 30 years at the baseline examination from March 25, 1985, to June 7, 1986 (year 0). Follow-up examinations were conducted at year 2 (response rate, 91%), year 5 (86%), year 7 (81%), year 10 (79%), year 15 (74%), year 20 (72%), year 25 (72%), and year 30 (71%). Among the 3498 participants who attended the CARDIA year 25 examination (June 1, 2010, to August 31, 2011), we included persons with documented systolic and diastolic BP measurements in 8 examinations who were free of cardiovascular disease, resulting in an analytic sample of 2400 participants. In an additional sensitivity analysis to further test the robustness of the findings, we examined a subset of participants (n = 1698) who never used BP-lowering medications and had a mean BP level of less than 130/80 mm Hg.
Blood pressure was measured using standardized protocols across all study examinations. Participants were asked to fast and abstain from heavy physical activity for at least 12 hours before evaluation. Participants were then seated in a quiet room for 5 minutes, and trained staff obtained 3 readings from the brachial artery. The mean of the second and third BP readings was used. In the years 0 to 15 examinations, a random zero sphygmomanometer was used. However, for the years 20 and 25 examinations, because of concerns about mercury contained in the apparatus, automated oscillometric BP monitors were used. A calibration study was performed, and values standardized to the sphygmomanometric measures were used for CARDIA study years 20 and 25 BP measurements to limit measurement bias.6 Visit-to-visit systolic and diastolic BP variability from years 0 to 25 was the exposure of interest in the present study. Variability measures used were SD, average real variability (ARV), and variability independent of the mean (VIM).
Average real variability was calculated using the following formula:
where N denotes the number of BP measurements in the analytic data set (N = 8) and ARV represents the absolute difference between successive BP measurements. In contrast to SD, ARV takes the order of BP measurements into account when quantifying BP variability. Because both SD and ARV are partly dependent on changes in systemic BP levels, biases from systemic changes in BP levels may be present even after adjustments in regression models.7 We therefore also determined the VIM,8 which was calculated as the SD divided by the within-individual mean to the power p and multiplied by the mean value of the systolic or diastolic BP in the cohort to the power p.
The CARDIA study risk factor assessments have been described previously.9,10 In summary, covariates were obtained using a combination of self-reported standardized questionnaires, clinical data, and laboratory data. Weight and height were measured with standardized procedures and used to calculate body mass index (BMI) (calculated as weight in kilograms divided by height in meters squared). Smoking status was self-reported using standardized questions and documented as never, former, or current smokers. Antihypertensive medication use was assessed by self-report and review of medical records. A modified version of the Minnesota Leisure Time Physical Activity questionnaire was used to capture moderate to vigorous physical activity. Fasting samples were collected using standardized protocols. Plasma cholesterol levels were assessed using enzymatic assays by Northwest Lipids Research Laboratory.
The year 25 examination echocardiographic protocol has been reported previously.11,12 In brief, the year 25 examination followed guidelines of the American Society of Echocardiography.13 Trained sonographers used a commercially available ultrasonographic system (Artida; Toshiba Medical Systems) with a 1.8- to 4.2-MHz phased-array transducer to acquire images from all field sites, and experienced readers at a core reading center interpreted digitized images. Left-ventricular (LV) mass was derived from the Devereux formula. Left-atrial and LV volumes were estimated from apical views. Left-ventricular ejection fraction (LVEF) was calculated from LV end-diastolic and end-systolic volumes. Tissue Doppler was used to assess the peak early diastolic velocity (E) and early peak diastolic mitral annular velocity (é). The mean of the septal and lateral mitral annular velocities was used. A more positive E/é (mitral inflow velocity to early diastolic mitral annular velocity) represents higher LV filling pressures, and a more negative é denotes impaired LV relaxation. Higher LV filling pressures and impairments in LV relaxation are representative of worse diastolic function. Speckle tracking echocardiography was performed (Advanced Cardiology Package 2D Wall Motion Tracking, version 3.0; Toshiba Medical Systems Co). Global longitudinal strain was defined as the peak systolic change in the segment length relative to the length at end-diastole and was represented as a percentage. Global longitudinal strain has increased prognostic value over LVEF, and a more positive global longitudinal strain represents worse myocardial deformation (systolic function).14 The year 25 echocardiographic examination had excellent reproducibility.11
Data analysis was conducted from June 7, 1986, to August 31, 2011. Continuous variables are presented as mean (SD). Categorical variables are reported as frequency and proportion. Pearson correlation coefficients were computed for BP variability measures. The primary goal of the analysis was to assess the association between visit-to-visit BP variability measures and echocardiographic outcomes. We therefore used multivariable linear regression models to assess visit-to-visit variability in systolic and diastolic BP (explanatory variables) and echocardiographic outcomes (dependent variable separate models for each: LV mass, left-atrial volume, LVEF, global longitudinal strain, E/é, and é). Models were adjusted for year 25 covariates: age, sex, race, educational level, BMI, smoking status, fasting glucose level, high-density lipoprotein cholesterol level, total cholesterol level, physical activity, and antihypertensive medication use (year 0 to year 25). In models assessing BP variability using SD and ARV, we additionally adjusted for mean BP level. To examine the possibility that BP-lowering medications and hypertension can have residual effects even after statistical adjustment, we carried out a sensitivity analysis in a subset of study participants who had a mean BP level less than 130/80 mm Hg and never used any BP medications in any study visit. Racial or sex differences in the association between BP variability and echocardiographic outcomes were assessed. Collinearity was not detected based on a variance inflation factor of less than 2.5. A 2-tailed P value <.05 was considered to be statistically significant. All analysis was performed using Stata, version 15.1 (StataCorp).
Population characteristics are presented in Table 1. Mean (SD) age at the year 25 examination was 50.4 (3.6) years; 1024 participants were men (42.7%), and 976 were African American (40.7%). At the year 25 examination, the mean (SD) BMI was 29.9 (7.1) and 319 participants (13.3%) were current smokers. Mean (SD) systolic BP was 117.9 (15.3) mm Hg and mean (SD) diastolic BP was 73.3 (10.9) mm Hg. Mean (SD) visit-to-visit systolic BP VIM was 8.4 (3.4), whereas mean (SD) diastolic BP VIM was 7.2 (2.8). The distribution of systolic and diastolic BP is illustrated in a histogram (eFigure in the Supplement). Systolic BP SD and ARV were moderately correlated (Pearson r = 0.59). Both systolic BP VIM and mean systolic BP (Pearson r = 0.03) and diastolic BP VIM and mean diastolic BP (Pearson r = 0.002) were poorly correlated. Compared with the excluded participants in the overall cohort, the analytic sample was more likely to be men, to be African American, and to have lower systolic BP, and there was no statistically significant difference in diastolic BP (eTable 1 in the Supplement).
In the Figure, linear prediction plots with 95% CIs show the association between visit-to-visit systolic BP variability and echocardiographic measures. As given in Table 2, per 1-SD increment, greater visit-to-visit systolic BP VIM was associated with higher LV mass index (β [SE], 2.66 [0.4]; P < .001), worse diastolic function (é) (β [SE], −0.40 [0.1]; P < .001), higher LV filling pressures (E/é) (β [SE], 0.37 [0.1]; P < .001), and worse global longitudinal strain (β [SE], 0.17 [0.1]; P = .002). Similarly, as reported in Table 3, greater visit-to-visit diastolic BP VIM was associated with higher LV mass index (β [SE], 3.21 [0.5]; P < .001), worse diastolic function (é: β [SE], −0.24 [0.1]; P < .001; E/é: β [SE], 0.23 [0.1]; P < .001), and worse global longitudinal strain (β [SE], 0.13 [0.1]; P = .02). We did not observe any association between visit-to-visit BP variability and either LVEF or left atrial volume. The findings remained consistent when 2 other BP variability measures were used (SD and ARV). There were no statistically significant racial or sex differences in the associations between BP variability and echocardiographic outcomes. We conducted a sensitivity analysis in a subset of participants who never used BP medications and had a mean BP level less than 130/80 mm Hg (n = 1698) (eTable 2 and eTable 3 in the Supplement). In this subset, the magnitude of associations between visit-to-visit systolic and diastolic BP variability and echocardiographic outcomes was attenuated, but the findings were consistent overall. In a sensitivity analysis, visit-to-visit systolic and diastolic BP variability was also associated with LV hypertrophy and diastolic dysfunction (eTable 4 in the Supplement).
In this community-based sample of young adults followed up for 25 years, greater visit-to-visit systolic and diastolic BP variability was associated with unfavorable alteration in myocardial structure and function independent of the mean BP levels. Our findings remained consistent even among a subset of participants who had a mean BP level less than 130/80 mm Hg and had never used BP-lowering medications. Our results are relevant from a public health perspective because even modest population-level increases in echocardiographic indices may have prognostic value for adverse outcomes.4,15 Our study observations lend support to a distinct association of visit-to-visit BP variability with risk of heart failure because of the associations with adverse cardiac remodeling.
To our knowledge, this study is the first to comprehensively investigate the association between long-term BP variability and cardiac remodeling. In 3 small studies that focused on short-term systolic BP variability (≤24 hours), higher systolic BP variability was associated with adverse LV alterations.16-18 Taking this finding a step further, Sega et al19 showed that the noncyclical (ie, erratic) component of 24-hour BP variability, but not the routine cyclical BP component, was associated with LV mass. It is plausible that the erratic component of 24-hour BP variability is more closely aligned with long-term BP variability than routine cyclical changes in BP. Our observations add to the scientific literature by providing data on long-term BP variability in a community-based sample with detailed assessment of cardiac structure and function. In the present study, we noted associations of visit-to-visit systolic and diastolic BP variability with higher global longitudinal strain but not with LVEF. Global longitudinal strain is a marker of systolic function that reflects longitudinal myocardial deformation and has been established to provide incremental prognostic value for outcomes compared with LVEF.14,20,21 In our analysis, we found no association between visit-to-visit systolic BP variability and left atrial volume. However, we noted associations between greater visit-to-visit systolic and diastolic BP variability and impaired LV relaxation and higher LV filling pressures, which may precede LA remodeling. Our study observations are notable because cardiac structural and functional alterations can antedate the risk of heart failure22,23 and other cardiovascular disease outcomes in general.15,20,24,25
There is accumulating evidence of pathophysiologic pathways that link greater BP variability with cardiac remodeling. Long-term, visit-to-visit BP variability is associated with factors that are also involved in cardiac remodeling, such as environmental, neurohormonal, and genetic factors, and cardiovascular disease risk burden.26-29 Studies have shown that BP variability is associated with endothelial dysfunction, coronary atheroma progression,30,31 and arterial remodeling,32-35 all of which are implicated in cardiac remodeling. In a murine study, mice that had no hypertension but had spontaneous variations in BP were reported to have desensitization of the β2-adrenoceptor, which is important for β-adrenoceptor agonist–induced relaxation in the large arteries.36 By exerting cyclical mechanical stress on the arterial wall, BP variability may facilitate vascular smooth muscle cell proliferation, extracellular matrix deposition, fibrotic changes, and microvascular alterations.37,38 In addition, there is increasing evidence of a possible bidirectional positive feedback loop wherein arterial stiffness indicates the probability of BP variability and increased BP variability can lead to greater arterial stiffness.33,34 Changes in arterial material properties exert a pressure load on the LV that may induce compensatory LV remodeling to normalize wall tension.9,39,40 Blood pressure variability may also have direct effects on cardiac structure and function.41 In another experimental study that performed sinoatrial denervation to produce increases in the variability of BP without changes in mean BP, greater BP variability was associated with perivascular macrophage infiltration, myocardial fibrosis, cardiac hypertrophy, and LV systolic dysfunction.41 Collectively, these reports are supportive of a potential association between BP variability and cardiac structure and function.
Our study results are strengthened by the use of standardized BP protocols and contemporary echocardiographic assessment in a large, biracial community-based sample; long-term study follow-up with good cohort retention; and the ability to account for a broad range of confounding factors.
Limitations of the study include the possibility of additional variability in periods between study visits. The optimal number of visits to assess BP variability is uncertain; however, it has been reported that at least 6 visits may be sufficient to estimate the cardiovascular disease risk associated with long-term BP variability.1 Ambulatory BP readings were not investigated in the present study; thus, our findings may not be generalizable to this context. Antihypertension medication use, drug dose or type, and medication nonadherence may be associated with BP variability; however, our findings remained consistent even among a subset of participants who never used BP-lowering medications. Because this was an observational study and our analysis did not examine change in echocardiographic indices over time, reverse causation and causal inferences cannot be determined. We attempted to account for confounding factors, but residual confounding from measured or unmeasured variables remains a possibility. Owing to the limited mid-life heart failure events from year 25 in this young cohort, we were not able to assess associations with heart failure. Although excluded CARDIA study participants had a higher comorbidity burden, the analytic sample in the present study represented a diverse cohort with a wide spectrum of cardiovascular risk factors. To the extent that the exclusion of high-risk participants likely biases toward the null, associations between BP variability and myocardial structure and function may be underestimated.
Greater visit-to-visit systolic and diastolic BP variability in young adulthood appears to be associated with unfavorable subclinical alterations in myocardial structure and function by middle age independent of the association with mean BP levels. Our study observations support a distinct association of long-term BP variability with risk of heart failure because of the association of subclinical echocardiographic features with heart failure.
Accepted for Publication: February 25, 2020.
Corresponding Author: João A. C. Lima, MD, MBA, Division of Cardiology, Department of Medicine, Johns Hopkins University School of Medicine, 600 N Wolfe St, Blalock 524, Baltimore, MD 21287 (firstname.lastname@example.org).
Published Online: April 15, 2020. doi:10.1001/jamacardio.2020.0799
Author Contributions: Drs Nwabuo and Lima had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Nwabuo, Yano, Appiah, Aghaji, Lloyd-Jones, Lima.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Nwabuo, Vasconcellos, Aghaji, Lima.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Nwabuo, Aghaji, Schreiner.
Obtained funding: Schreiner, Lloyd-Jones, Lima.
Administrative, technical, or material support: Lima.
Supervision: Yano, Lloyd-Jones, Lima
Conflict of Interest Disclosures: None reported.
Funding/Support: The Coronary Artery Risk Development in Young Adults study is conducted and supported by the National Heart, Lung, and Blood Institute (NHLBI) in collaboration with the University of Alabama at Birmingham (grants HHSN268201800005I and HHSN268201800007I), Northwestern University (grant HHSN268201800003I), University of Minnesota (grant HHSN268201800006I), and Kaiser Foundation Research Institute (grant HHSN268201800004I). Drs Shah and Murthy are supported by 1R01HL136685 from the NHLBI and grant 1R01AG059729 from the National Institute on Aging.
Role of the Funder/Sponsor: The sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Meeting Presentation: This study was presented at the ACC 2020: American College of Cardiology 69th Scientific Session and Expo; March 29, 2020 (online).
Create a personal account or sign in to: