Visit-to-Visit Blood Pressure Variability—What is the current challenge?

Variability is the law of life. The concept of entropy and the second law of thermodynamics suggest that systems naturally progress from order to disorder. Conversely, most biological systems appear to be homeostatic.1 However, it is obvious that biological signals, such as blood pressure (BP) and heart rate, vary in complex and irregular ways even at a steady-state condition. This is because physiological functions are a result of continuous, dynamic interactions between multiple neural, hormonal, mechanical, and local subcellular biochemical control systems.2 These physiological functions are easily altered by environmental factors, including temperature, diet, mental or physical stress, and sleep. Particularly, aging, characterized by progressive decline in tissue and organ function, has a dramatic effect on whole-body physiology. Therefore, age is a robust contributor to BP variability, both short-term (e.g., beat-to-beat and daily) and long-term (e.g., day-by-day and across outpatient visits).3,4 In clinical practice with older adults, extreme values of BP measured during outpatient visits are often discarded because of the “white-coat effect,” and variability of BP across outpatient visits (i.e., visit-to-visit BP variability) is dismissed as a random fluctuation. When considering multiple BP measurements, patients are prone to regard the lowest BP measurement as their true BP.5 However, visit-to-visit BP variability is reproducible (not random),6 and accumulative evidence links high visit-to-visit BP variability to adverse outcomes in older adults.3,4,7–9 Researchers and clinicians face 3 significant challenges in deciding how best to utilize measures of BP variability: Developing an effective strategy to implement research findings into health policy and practice; Determining which visit-to-visit BP variability measurements should be used to identify those at a greater risk of adverse outcomes, and how different time intervals between BP measurements and the number of visits could influence the clinical relevance of long-term BP variability; and Determining whether visit-to-visit BP variability is a modifiable risk factor. In the current issue of American Journal of Hypertension, Wu et al., have provided evidence regarding the second issue stated above, by showing a new potential clinical implication of visit-to-visit BP variability in community-based older adults.10 These investigators used data from the Health, Aging, and Body Composition Study that recruited well-functioning older adults 70–79 years of age (n = 1,877; mean age 73 years; 53–55% women; and 33–41% African Americans) in 1997–1998. Office BP was measured annually over 5 years, with follow-up visits conducted over an average of 8.5 years. The primary exposures were systolic BP (SBP)/diastolic BP (DBP) variability, and primary outcomes included all-cause and cardiovascular disease (CVD) mortality. With adjustments for covariates, results from Cox models suggested that higher visit-to-visit DBP variability—but not SBP variability—was associated with higher all-cause and CVD mortality, independent of mean BP over visits. Results were generally similar across subgroups defined by the presence or absence of antihypertensive medication use. The novelty in this study includes its assessments of visit-to-visit BP variability and distinct BP components (i.e., SBP, DBP, and pulse pressure defined as SBP minus DBP) that were variously associated with distinct adverse outcomes (i.e., all-cause mortality, CVD morality, myocardial infarction, and stroke). The SD, coefficient of variation, the maximum and minimum BP difference, and average real variability (ARV) across multiple visits have been widely used to describe visit-to-visit BP variability in previous studies.3–9,11–13Figure 1 (left) shows one example of individual follow-up data of SBP across 5 visits. Generally, SD, coefficient of variation, and maximum and minimum BP difference are strongly correlated (Pearson’s r > 0.90).14 Despite the strong correlation between ARV and SD or coefficient of variation (Pearson’s r = 0.7–0.8),14 their clinical implication may not be identical. For example, ARV weights for the between-reading time intervals and takes the order of the BP measurements into account. Conversely, SD is influenced by outliers or extreme BP values, which may be triggered by lifetime adverse events, including illness and environmental stressors, and natural disaster.3,4 Of concern is the possibility that these parameters are partially dependent on the overall BP level and change in mean BP levels over time; distinguishing BP variability from systemic changes in BP level over time (e.g., slope) is difficult. The issue may not be resolved even if mean BP level over visits is used as an adjustment factor. Therefore, Wu et al., used residual SD, defined as visit-to-visit BP variability calculated as the root-mean-square error of the residuals (i.e., differences between observed BP and predicted BP) from the regression (Figure 1, right). This measurement is less influenced by BP level change over time compared with SD. However, it should be noted that the estimation of residual SD is based on the assumption that a participant’s BP increases “linearity” over time. BP variability independent of the mean (VIM) is another visit-to-visit BP variability measurement that has no correlation with mean BP level over visits.4 VIM is calculated as the SD divided by the mean to the power x and multiplied by the population mean to the power x. The power x is obtained by fitting a curve through a plot of SD against mean using the model SD = a times meanx, where x was derived by nonlinear regression analysis as implemented in the SAS PROC NLIN procedure. However, the estimation of VIM is derived from the distribution of BP within each cohort, and thus the value itself cannot be compared across populations. Adjustments for cumulative exposure to BP (Figure 1, left)11 in models, including SD and ARV as the primary exposure, would overcome these limitations, since cumulative exposure to BP reflects not only mean BP level but also systemic change in BP level over time.15 Illustration of SD, CV, difference between maximum and minimum BP, ARV, and residual SD over visits. The figures show one example of individual follow-up data of SBP across 5 visits. Left Figure: The absolute differences of SBP between successive SBP measurements are shown as Δ1 − Δ4. For example, Δ1 represents the absolute difference in SBP between visit 1 and visit 2 SBPs. ARVSBP is calculated as (Δ1 + Δ2 + Δ3 + Δ4)/4. MMD is calculated as maximum SBP minus minimum SBP from visit 1 to visit 5. Mean SBP and SDSBP over visits is calculated from 5 SBP measurements (visit 1 − visit 5) for each individual, and CV is calculated as SD/mean SBP over visits. The average SBP between successive BP measurements is shown as A1 − A4. Cumulative exposure to SBP will be calculated as (A1×2 years + A2×3 years + A3×2 years + A4×3 years) and is shown by the blue dotted area, representing mm Hg × years. Right Figure: Intraindividual SBP change over time (i.e., slope) is defined as the beta coefficient for the linear regression over the 5 SBP measurements. Intraindividual variability is defined as the SD of the residuals (r1, r2, r3, r4, and r5) from the linear regression of the 5 SBP measurements (residual-mean-square error). Abbreviations: ARV, average real variability; CV, coefficient of variation; BP, systolic blood pressure; MMD, maximum and minimum BP difference. Currently, there is no consensus on a gold-standard approach to measure visit-to-visit BP variability. Consequently, the clinical implementation of visit-to-visit BP variability is a great challenge. Without question, the primary goal of visit-to-visit BP variability research should be to determine its potential causal effects, with careful specification of all potential confounders. For this purpose, residual SD and VIM appear to be an ideal approach based upon the relevant literature to date. The next goal is to tackle the issue of implementation and application to clinical-decision making. For example, if the predictive ability of incident CVD is similar between residual SD (or VIM) and ARV (or SD)7,9 and these predictive abilities are not complementary, ARV (or SD) may be easier measures for clinicians to obtain. Translational research cannot be completed unless findings are used to change clinical practice, and thus these 2 goals need to be considered concomitantly in future BP variability research. Different BP components reflect distinctive hemodynamics and pathophysiology.16,17 SBP tends to rise with age, whereas DBP tends to drop after age 55–60 as a consequence of atherosclerosis and reduced elasticity/compliance of large arteries.18 Pulse pressure in older adults is a measure of pulsatile pressure load and is most representative of large arterial stiffness.16,17 Wu et al., demonstrated that higher DBP variability and an excess (≥10mm Hg) drop in DBP between 2 consecutive visits were associated with a higher risk of all-cause and CVD mortality. Why was no correlation found between SBP and outcomes? One potential explanation is that a drop in DBP may reduce organ blood flow (the heart in particular, since it is perfused during diastole). If so, why was DBP variability or an excess drop in DBP not associated with incident myocardial infarction or stroke events? Is visit-to-visit DBP variability in older adults a surrogate for homeostasis disruption or frailty instead of a risk factor?19 Excess BP variability appears to enhance vascular damage and lead to periods of organ hypoperfusion. This may be true when “short-term” (e.g., beat-to-beat) BP variability is high. Could this phenomenon also happen when visit-to-visit “long-term” BP variability is high? Generally, short-term and long-term BP variability are weakly correlated,3,4,12 suggesting that their pathophysiology may not be identical. Visit-to-visit BP variability may integrate complex interactions between extrinsic environmental and behavioral factors over a lifetime (e.g., nutrition, sleep, and psychosocial factors) and intrinsic cardiovascular homeostasis.3,4 Therefore, whether visit-to-visit BP variability in older adults is a contributor to adverse outcomes or merely an epiphenomenon of certain pathological conditions remains uncertain in Wu’s study. Unraveling the mechanistic links between high visit-to-visit BP variability and adverse outcomes will require further investigation. The resulting knowledge may translate to new therapeutic targets directing at reducing BP variability (not necessarily via antihypertensive agents)20 and ultimately improve outcomes for individuals. Lastly, it should be noted that BPs in the Health, Aging, and Body Composition Study were collected using standardized protocols by trained research staff. DBP is generally more difficult than SBP to measure precisely. Therefore, the clinical relevance of visit-to-visit DBP variability “in real-world medical practice” merits further investigation.