Speed of Heart Rate Recovery in Response to Orthostatic ChallengeNovelty and Significance
Rationale: Speed of heart rate recovery (HRR) may serve as an important biomarker of aging and mortality.
Objective: To examine whether the speed of HRR after an orthostatic maneuver (ie, active stand from supine position) predicts mortality.
Methods and Results: A longitudinal cohort study involving a nationally representative sample of community-dwelling older individuals aged ≥50 years. A total of 4475 participants completed an active stand at baseline as part of a detailed clinic-based cardiovascular assessment. Beat-to-beat heart rate and blood pressure responses to standing were measured during a 2-minute window using a finometer and binned in 10-s intervals. We modeled HRR to the stand by age group, cardiovascular disease burden, and mortality status using a random effects model. Mortality status during a mean follow-up duration of 4.3 years served as the primary end point (n=138). Speed of HRR in the immediate 20 s after standing was a strong predictor of mortality. A 1-bpm slower HRR between 10 and 20 s after standing increased the hazard of mortality by 6% controlling for established risk factors. A clear dose–response relationship was evident. Sixty-nine participants in the slowest HRR quartile died during the observation period compared with 14 participants in the fastest HRR quartile. Participants in the slowest recovery quartile were 2.3× more likely to die compared with those in the fastest recovery quartile.
Conclusions: Speed of orthostatic HRR predicts mortality and may aid clinical decision making. Attenuated orthostatic HRR may reflect dysregulation of the parasympathetic branch of the autonomic nervous system.
Much recent work has focused on the prognostic value of heart rate recovery (HRR) postexercise as a risk factor for cardiovascular disease (CVD) and mortality.1–5 Vagal reactivation plays an integral role in modulating the rate at which the heart rate recovers after exercise, especially during the first 30 s6 and ≤2 minutes.3 A reduction in vagal tone and an increase in activity of the sympathetic nervous system are associated consistently with an increased risk of cardiovascular events, including sudden death4,7,8 and with all-cause mortality. More recently, the vagus nerve has also been shown to have a reflexive role in modulating proinflammatory signaling,9,10 which may, in part, contribute to the association with CVD and mortality.
Editorial, see p 582
In This Issue, see p 573
The autonomic nervous system plays a central role in regulation of cardiovascular and humoral responses to orthostasis. Orthostasis evokes a rapid physiological response involving the co-ordinated action of several systems, including the skeletal-muscle pump and arterial and cardiopulmonary baroreflexes.11 The orthostatic response reflects a balance between cardiac output and total peripheral resistance modulated by the autonomic nervous system and baroreflexes and most commonly measured using changes in heart rate and blood pressure. When resting in the supine position, venous and arterial reservoirs are at the same height, but standing up reduces venous return by displacing ≈500/700 mL (10 mL/kg) of central blood into the peripheral system.
Heart rate increases rapidly in the first few seconds after standing to counteract the gravitational forces acting on blood pressure that propels blood toward the lower extremities (Figure 1). The initial surge in heart rate that occurs in the first few seconds after standing results from abrupt inhibition of vagal activity.12–14 The peak heart rate that is reached at about 10 s after standing is a product of vagal inhibition and (slower acting) sympathetic systems acting in concert. Heart rate declines rapidly after this point as a result of rebounding arterial pressure.13 There is a particularly steep drop in heart rate between 10 and 20 s and an age gradient with heart rate and blood pressure responses.15
With a few exceptions,2 most of the studies examining HRR in response to exercise have been conducted in symptomatic, clinically referred samples so the ecological validity of these findings for the general population remains questionable. Furthermore, there is no standard protocol for measuring HRR.16,17 Recovery values have been measured at time intervals ranging from 1 to 5 minutes after the cessation of treadmill exercise, and what constitutes a slow HRR has varied from study to study. Few studies have explored the hemodynamics of the HRR during the time period when the speed of recovery toward baseline is at its most pronounced; that is, in the immediate 30 s after the cessation of exercise.18 Finally, HRR to orthostasis has not been widely studied, despite the fact that the results of at least 2 studies suggest that shifts in cardiac autonomic balance in response to standing may have promise as predictors of mortality.19,20
Given that standing can be easily performed by anyone who is functionally mobile and is a potent cardiovascular stressor that demands the full capabilities of the reflexes that govern cardiovascular function,21 we examined the predictive value of HRR to standing in a large adult population study and further describe a novel measure of autonomic dysregulation.
The Irish Longitudinal Study on Ageing (TILDA) is a population-based nationally representative cohort study of aging in the Republic of Ireland that measures continuous noninvasive beat-to-beat heart rate and blood pressure responses to orthostatic change in community-dwelling older individuals aged ≥50 years as part of a comprehensive multidisciplinary assessment.22 In this article, we describe the development and validation of a new biomarker of cardiovascular aging derived from the active stand procedure, specifically the speed of HRR between 10 and 20 s after standing, and show that this parameter has clinical relevance as a marker of aging, CVD, and all-cause mortality.
Study Design and Participants
TILDA is a large prospective cohort study examining the social, economic, and health circumstances of 8175 community-dwelling older adults, aged ≥50 years, resident in the Republic of Ireland. The sample was generated using a 2-stage clustered sampling process and the Irish Geodirectory as the sampling frame. The Irish Geodirectory is a comprehensive listing of all addresses in the Republic of Ireland, which is compiled by the national post service and ordnance survey Ireland. The primary sampling units were 640 geographic regions selected by random selection, stratified on proportion of head of households in the professional class, proportion of the population aged ≥65 years, and geographical location. The second stage involved the selection of a random sample of 40 addresses from within each primary sampling unit, resulting in an initial sample of 25 600 addresses. Addresses were then assessed for eligibility, and members of eligible households aged ≥50 years were canvassed to participate. Consequently, the response rate was defined as the proportion of sampled households, including an eligible participant from whom an interview was successfully obtained. A response rate of 62.0% was achieved at the household level.23 The baseline survey (wave 1) occurred in 2009/2011.
Respondents completed a computer-assisted personal interview (n=8175) in the home. All participants were subsequently invited to undergo a detailed clinic-based health assessment in 1 of 2 national centers using trained nursing staff. A total of 5035 people attended the health center assessment at wave 1. One hundred fifteen individuals were unable to complete the active stand, and data for a further 445 individuals were excluded because of poor signal quality, incomplete data, or poor compliance with protocol.15 This left a final total of 4475 people who completed the active stand procedure. Online Figure I presents the flow diagram for study participation.
Ethical approval for the study was obtained from the Trinity College Dublin Research Ethics Committee. Signed informed consent was obtained from all participants.
Heart Rate and Blood Pressure Measurement
A detailed description of the active stand protocol used in TILDA is available elsewhere.24 Briefly, participants who attended the health center completed an active stand from a supine position as part of a detailed cardiovascular health assessment. A pressure cuff was applied to the finger of each participant to measure their phasic blood pressure. Participants rested comfortably in the supine position for 10 minutes before performing the stand in a silent room with an ambient temperature ranging between 21 and 23°C. Participants were asked to stand in a timely manner (<5 s) under the supervision of a nurse and were assisted to stand if this proved necessary.
The zero time point for each individual was set by the clinical nurse at the point where the participant began to rise from the supine position. Beat-to-beat variability in heart rate and blood pressure during the stand was captured using noninvasive digital photoplethysmography (Finometer, Finapres Medical Systems, Arnhem, The Netherlands). Data were corrected for hydrostatic changes in finger pressure because of standing using the Height Correction Unit used by the Finometer device. The following parameters were extracted:
The supine baseline values of heart rate (HR), systolic blood pressure (SBP), and diastolic blood pressure (DBP) occurring 60 s before standing.
Recovery values for heart rate, SBP, and DBP at 10-s time intervals between 10 and 110 s after the stand. The full beat-to-beat data traces were filtered using a nonstationary moving average filter. Recovery values at each 10-s time interval represented a moving average ±2.5 s around that time point. These values are denoted HR(t), SBP(t), and DBP(t), where t is time in seconds after standing and takes on values of 10 to 110 s.
Difference from baseline measures were obtained by subtracting values of HR(t) at each time point from the baseline resting heart rate (supine baseline values of heart rate). These values are denoted ΔHR(t). This process was repeated for SBP and DBP. These values are denoted ΔSBP(t) and ΔDBP(t).
End Points: All-Cause Mortality
Mortality status was established through attempts at contact with participants at wave 2 (≈2 years after baseline) and wave 3 (≈4 years after baseline). In total, 549 of the 8175 individuals or 6.7% of the sample who were initially recruited were confirmed as deceased by up to a maximum of 6-year follow-up. Of the 4475 individuals who completed the active stand at wave 1, 141 were confirmed as dead as of 18 December, 2015 (ie, the last interview date for wave 3 data collection), which is an effective mortality rate of 3.2%.
Time of death was available for 72% (n=102/141) of the deceased from an end of life interview that was conducted with the respondent’s surviving kin. Time of death was unavailable for the remaining 28% of the sample because TILDA allows a period of 6 months to elapse before attempting to conduct an end of life interview with the spouse/family of recently deceased cohort members. In these instances, time of death was determined if a family member/spouse of the deceased informed the TILDA fieldwork team of the date of death. In the remaining cases of confirmed deaths, time of death was established using a database operated by funeral directors in Ireland to notify deaths. Respondents were identified on the basis of a name and address match. There were 3 remaining cases where date of death could not be determined. In these instances, we imputed date of death as half of the mean between-wave interval after the date of last contact.
It was not possible to establish whether all participants at wave 1 were either deceased or still living by wave 3. Of the sample of 8175 completing wave 1 and the subset of 4475 completing the active stand, respectively, 764 (9.3%) and 166 (3.7%) were either lost to follow-up or refused to participate further by wave 2 and did not participate at wave 2 and wave 3. The surveillance interval was not constant for all participants because of unequal intervals between interviews, loss to follow-up, and mortality itself. This interval ranged from 1 day to 5.8 years, with an average of 4.3 years. One hundred thirty-eight individuals who completed the stand at wave 1 and were confirmed as subsequently deceased had complete information on all covariates.
Primary Predictor Variable
We initially explored variation in HRR across the stand using 3 broad age groups: 50 to 59, 60 to 69, and ≥70 years. Preliminary investigation of the data revealed that the speed of HRR in the early part of the stand differentiated strongly between age groups; specifically the speed of HRR between 10 and 20 s after standing. We extracted an additional parameter to describe the speed of HRR during this time window (described in the statistical analysis section below), and it served as the primary predictor variable in the analysis.
The covariates were chosen based on their association with mortality and speed of HRR in the literature. All of the covariates were measured at baseline during the wave 1 sweep of data collection. In addition to age, sex, resting HR (bpm), and resting SBP (mm Hg), we control for the use of anticardiovascular medications. The international nonproprietary name for any regularly taken medications was assigned and coded using anatomic therapeutic classification codes. Cardiovascular medications were antiadrenergics (C02), diuretics (C03), β-blockers (C07), calcium channel blockers (C08), and angiotensin-converting enzyme inhibitors (C09). Medical history including pre-existing doctor-diagnosed CVDs that represent hard end points (angina, heart attack, congestive heart failure, stroke, and transient ischemic attack) were ascertained during the household interview. Participants with atrial fibrillation were identified as such if they self-reported having an abnormal heart rhythm and this was confirmed from the ECG recording. These data were then pooled to create a 3-level CVD disease measure: CVD free, 1 CVD, and ≥2 CVDs, for use in the analysis.
We also include controls for several specific comorbidities that could affect exercise and exertion levels required when actively standing from the supine position. Having ever received a doctor diagnosis of cancer, lung disease, or diabetes mellitus is represented by a series of binary variables in the analysis. Limitations in activities of daily living (ADLs) and instrumental activities of daily living (IADLs) are included as a proxy for the participant’s general physical condition. ADLs included difficulties with (1) dressing, (2) walking across a room, (3) bathing or showering, (4) eating, such as cutting up food, (5) getting in or out of bed, and (6) using the toilet. IADLs included (1) difficulties in preparing a hot meal, (2) doing household chores, (3) shopping for groceries, (4) making telephone calls, (5) taking medications, and (6) managing money. We summed the number of conditions separately with respect to ADLs and IADLs, and the count of these conditions is used in the analysis.
Lifestyle factors included smoking history, which is represented as a 3-level variable: never smoked, past smoker, current smoker; body mass index (measured weight/measured height m2); and serum lipids. Body mass index was measured at the clinic visit by trained nursing staff using scientifically approved and calibrated measuring equipment. Participants also provided a blood sample during the course of the health assessment, and these were sent for immediate analysis to derive a detailed lipid profile, which included high-density lipoprotein, low-density lipoprotein, and triglycerides. Finally, we also include a control for educational attainment, which is represented as a 3-level variable: primary, secondary, and tertiary education.
Repeated observations of HR at 10-s intervals within a cross-section allow treatment of the data as a panel (measurement occasions nested within individuals) and fitting of a random effects model using generalized least squares estimation. We explored age-related variation in HRR across the stand for different age groups (50–59, 60–69, and ≥70 years) controlling for sex, existing CVD, and use of cardiovascular medications by fitting the following model (Equation 1) to the data of n individuals, with an individual denoted by i at time j (tij) poststand:(1)
where i=(1,…,n), j=(1,…,11), and yij represents the difference in HR from baseline (ΔHR) at tij; α is the intercept; βj, the coefficient for each time point at the reference level of each covariate; Xi, a vector of individual-level covariates: age group (50–59, 60–69, and ≥70 years), sex, existing CVD (none, 1, and ≥2 CVDs), and use of cardiovascular medications (no and yes); and γ, the related row vector of coefficients. A cross-level interaction term between time (tij - level 1) and individual-level covariates (Xi - level 2) is given by tij Xi and where δ is the related row vector of coefficients. This allows HRR to vary over time by age group and by other covariate groups. The terms ui and eij are residuals representing an unobserved individual effect and an error term for individual i at time j, sampled from normal distributions with variances τ2 and σ2, respectively. The model, thus, contains 77 fixed-effects parameters and 2 random-effects parameters. The predictive margins at the means and the associated 95% confidence intervals for the cross-level (time×age group) interaction were derived and plotted.
Visual inspection of the resulting plots revealed that the speed of HRR in the early part of the stand (ie, initial 20 s) was the orthostatic feature that most clearly distinguished younger from older participants. Older people experienced a less vigorous increase in HR on standing and a slower recovery toward baseline between 10 and 20 s relative to those aged 50 to 59 years. Additional parameters were, therefore, extracted to represent the speed of HRR during this time frame by subtracting the difference from baseline value of HR at 10 s from the value at 20 s (Equation 2). This is equivalent to simply calculating the absolute difference in heart rate values between 10 and 20 s after standing.(2)
Differences in the rate of change (ie, slope) across age groups between 10 and 20 s after the stand was confirmed by performing significance tests for difference. Because it is clinically difficult to uncouple the speed of HRR from the BP response, parameters were also extracted to represent the speed of the SBP (Equation 3) and DBP recovery occurring at the same time point (Equation 4).(3)(4)
The bivariate association of each of the parameters extracted from the stand with participants’ age at baseline was examined using Spearman rank-order correlation coefficient. This analysis confirmed that the speed of HRR10s|20s was the parameter that was most strongly correlated with age. To check clinical relevance, we compared the speed of HRR10s|20s in those with and without CVD and examined its association with the probability of mortality during an average 4.3-year follow-up.
We also calculated the receiver operating characteristic curves predicting mortality in a series of separate univariate analyses with respect to each of the HR, SBP, and DBP parameters extracted from the stand, which was implemented using the ROCTAB (receiver operating characteristic)25 procedure in STATA.
We followed the American Heart Association’s recommendations for the evaluation of a novel risk marker26 to assess the prognostic value of the speed of HRR10s|20s. Cox proportional hazards models27 were fitted to the data to determine whether the speed of HRR10s|20s was associated with time to death (month and year of death) up to a maximum of 6 years after initial assessment. The crude model (model 1) estimated the impact of a 1-bpm change in the speed of HRR10s|20s on the hazard of mortality. Model 2 adjusted additionally for a range of covariates measured at baseline: (resting HR, resting SBP, sex, cardiovascular medications, CVDs, cancer, lung disease, diabetes mellitus, ADLs, IADLs, smoking status, body mass index, serum lipid profile, and educational status) to determine whether the speed of HRR10s|20s was independently associated with hazard of mortality in a multivariable model. Model 3 added age and an age squared term to the equation to determine whether the speed of HRR10s|20s predicted mortality independently of age.
Discrimination performance was assessed using Harrell C index and Somer D index. As described by Pennells et al,28 the C index estimates the probability of concordance between the predicted risk and the observed order of events for a randomly selected pair of participants. The D index estimates the mean log hazard ratio for the event of interest for a randomly selected pair of participants: one in the top half and one in the bottom half of the predicted risk distribution. We estimate first the model with the established risk markers and then we estimate the model with the established risk markers and the novel risk marker (HRR10s|20s) to determine whether it leads to an improvement in prediction. The predictive accuracy of the novel risk marker was further assessed using the net reclassification improvement index29 based on continuous predictions from a binary model predicting probability of mortality at the end of the surveillance period for each individual as described in the Methods section in the Online Data Supplement.
Missing Data and Nonresponse
Only 110 participants, 3 (2.3%) of the 141 confirmed deceased and 107 (2.5%) of the 4227 alive, at last contact had ≥1 missing covariates, so we report the results from the complete case analysis. SEs of estimates were adjusted to account for the clustered design effect and stratification, and data were weighted using survey weights to account for the fact that respondents who attended the health center were younger, better educated, and in better health.23
Mean age of the sample was 62.8 years (SD=9.2), 51.3% were female, and 35.5% were taking cardiovascular medications. Almost 12% of the sample had a doctor-diagnosed CVD. The mean HRR10s|20s was −5.78 bpm (SD=7.06). Table 1 describes the baseline characteristics of the sample and how they vary according to quartiles of HRR10s|20s. For example, mean age of those in the slowest HRR quartile was 67.8 years compared with a mean age of 57.7 years among those in the fastest HRR quartile. Similarly, 49.1% of those in the slowest HRR quartile were taking cardiovascular medications compared with 19.3% of those in the fastest HRR quartile. Figure 1 illustrates the mean unadjusted orthostatic HR, SBP, and DBP response to standing for the sample. It shows that HR increases rapidly in the first 10 s after standing and then declines quickly between 10 and 20 s. This pattern is more or less reversed with respect to SBP and DBP, which fall rapidly in the first 10 s but recover quickly between 10 and 20 s.
Table 2 summarizes the bivariate association between age and the array of HR, SBP, and DBP parameters extracted from the stand. The speed of HRR between 10 and 20 s (HRR10s|20s) after standing was the parameter that was most strongly correlated with age (r=0.40). It was more strongly correlated with age than any of the difference from baseline measures of HR, SBP, or DBP or indeed the speed of the SBP10s|20s (r=−0.16) and DBP10s|20s recovery (r=−0.23). The ΔHR10s was the only other variable extracted from the stand that correlated >0.30 (r=−0.33) with age.
Figure 2A through 2D shows that older participants were characterized by a slower HRR toward baseline in the immediate 20 s after standing compared with younger participants in the sample. The speed of HRR10s|20s was −8.21 bpm for those aged 50 to 59 years (Figure 2A), −5.25 bpm for those aged 60 to 69 years (Figure 2B), and −2.51 bpm for those aged ≥70 years (Figure 2C). Figure 2D shows the speed of recovery for all age groups simultaneously.
Figure 3A through 3D shows slower HRR10s|20s among those with higher CVD burden. Individuals who were free of CVD at wave 1 experienced a greater deceleration in heart rate between 10 and 20 s after standing (−6.30 bpm, Figure 3A) compared with those who had 1 CVD (−5.50 bpm, Figure 3B) or ≥2 CVDs (−4.26 bpm, Figure 3C). Figure 3D shows these relationships simultaneously. Formal statistical tests confirmed that the speed of HRR10s|20s was significantly faster among those who were CVD free compared with those who had ≥2 CVDs.
The speed of HRR10s|20s also distinguished those who completed the stand at baseline and had died during a mean 4.3-year follow-up. Figure 4A through 4C illustrates this relationship graphically showing that the average HRR10s|20s at wave 1 for those who subsequently died was −3.53 bpm compared with an average HRR10s|20s of −6.30 bpm for those who were still alive at follow-up. The results of the receiver operating characteristic analyses are presented in Online Table 1. The majority of the difference from baseline measures of HR, SBP, and DBP did not perform significantly better than chance in predicting mortality. Notably, the area under the curve was greatest for the speed of HRR10s|20s (area under the curve=0.69; 95% CI, 0.64–0.73), although comparable to the difference from baseline value of heart rate at 10 s: ΔHR10s (area under the curve=0.66; 95% CI, 0.62–0.71).
The speed of HRR10s|20s was related to all-cause mortality in univariable and multivariable Cox regression analyses. In the crude analysis (model 1), a 1 bpm slower HRR10s|20s was associated with a 10% increase (hazard ratio, 1.10; 95% CI, 1.08–1.13; P<0.001) in the hazard of all-cause mortality during a mean 4.3-year follow-up period. The association was robust to adjustment for a broad range of established risk markers measured at baseline (model 2), including resting HR, resting SBP, sex, use of cardiovascular medications, existing CVD burden, lung disease, cancer, diabetes mellitus, ADLs, IADLs, smoking, body mass index, serum lipids, and education (hazard ratio, 1.09; 95% CI, 1.06–1.13; P<0.001). The speed of HRR10s|20s remained a significant predictor of all-cause mortality even when adjusted additionally for age and age2 in model 3 (hazard ratio, 1.06; 95% CI, 1.03–1.10; P<0.001).
We observed small incremental gains in the C index (0.810 versus 0.816) and D index (0.620 versus 0.633) when we added speed of HRR10s|20s to the model containing the established risk factors predicting mortality. Net reclassification resulted in 78 deaths having greater predicted risk and 60 deaths having lower predicted risk in the model with HRR10s|20s compared to the model without. Similarly, 2503 nondeaths had lower predicted risk, and 1724 nondeaths had greater predicted risk in the model with HRR10s|20s compared with the model without. The continuous net reclassification improvement was 0.315 (95% CI, 0.147–0.483).
To facilitate exploration of dose–response effects, we calculated the hazard ratio and associated 95% confidence intervals for all-cause mortality according to quartiles of HRR10s|20s. The mean HRR10s|20s for those in the fastest recovery quartile was −15.31 bpm. By comparison, the mean HRR10s|20s for those in the slowest recovery quartile was positive (+2.11 bpm), which means that on average, these individuals experienced an increase in heart rate between 10 and 20 s after standing.
Kaplan–Meier survival curves according to quartiles of HRR10s|20s are depicted in Figure 5. Table 3 (model 1) shows that in the crude model, those in the slowest quartile of HRR10s|20s were 7.0× (95% CI, 3.7–13.4; P<0.001) more likely to experience mortality during a mean follow-up duration of 4.3 years compared with those in the fastest recovery quartile; and there was a clear dose–response relationship between the speed of HRR10s|20s and hazard of all-cause mortality. In the full multivariable adjusted model, those in the slowest recovery quartile remained 2.3× (95% CI, 1.1–4.5; P<0.05) more likely to experience all-cause mortality compared with those in the fastest recovery quartile.
In this large epidemiological study of aging, the speed of HRR between 10 and 20 s (HRR10s|20s) after standing was a clear risk marker for mortality during a mean 4.3-year follow-up. The reason why this slope parameter performs better than simply using difference from baseline measures at each time point is that it takes account of how far HR increases on standing and how quickly it recovers toward baseline. To some extent, we think it may be useful to conceive of the slope parameter as a measure of heart rate elasticity because it reflects both the capacity of the system to mount a vigorous response to cardiovascular challenge and to quickly re-establish homeostatic equilibrium. Viewed in this context, an attenuated HRR to the active stand may reflect dysregulation of the parasympathetic branch of the autonomic nervous system because parasympathetic inhibition is largely responsible for the initial surge in HR after the stand,13 whereas parasympathetic reactivation is thought to be responsible for the speed of HRR in the early phase of recovery.30,31 Imai et al30 found that pharmacological blockade of parasympathetic reactivation using atropine delayed cardiac heart rate deceleration, particularly in the initial 30 s of the postexercise recovery phase.
Consistent with expectations for a powerful risk marker, the speed of HRR10s|20s distinguishes between different age groups, demarcates those with existing CVD from those who are CVD free, and has predictive power as an indicator of all-cause mortality independent of other established risk markers. In the crude analysis, a 1-bpm slower HRR10s|20s was associated with a 10% increase in the hazard of all-cause mortality during a mean 4.3-year follow-up period. The risk of mortality increased as the speed of HRR10s|20s declined. An individual in the slowest quartile of HRR10s|20s was 7.0× more likely to die at follow-up compared with those in the fastest recovery quartile. These associations were robust to adjustment for established risk factors and remained significant even when adjusted for age, although the latter strongly attenuated the associations. What seems clear is that the speed of HRR10s|20s has clinical value as a marker of health and may help identify individuals at risk of mortality. Given that we have limited mortality data, the magnitude of the effect sizes would seem to indicate that we have identified a parameter from the stand that has important clinical relevance.
It should be acknowledged that the novel risk marker led to, at most, moderate gains in the discrimination and accuracy of prediction as assessed using measures of concordance (C index and D index) and net reclassification improvement. This is perhaps not unexpected given that, as a prospective epidemiological study of aging, the TILDA sample is so well characterized in terms of established risk markers. However, in the context of a clinical setting where all of these covariates are not readily available to the clinician, the speed of HRR10s|20s may serve as a useful adjunct to help guide clinical decision making.
Only a small number of cases who completed the stand (n=138) and had full information on covariates had died at follow-up, so the confidence intervals around the estimates are large. Despite this, the magnitude of the effect sizes support the use of orthostatic HRR as a robust risk marker. Second, information was not available about cause of death and the numbers would likely be too small to disaggregate and analyze by disease type. It is entirely possible that the predictive power of HRR10s|20s would be higher if we were predicting cardiovascular as opposed to all-cause mortality. Future research with this cohort will be directed toward this end, particularly if we are able to link individual-level clinical data to administrative sources of data (ie, National Death Registry). Finally, another criticism that could be leveled at the study is that a sizeable proportion of the sample did not attend the health center assessment, and a sizeable subset of individuals who did attend the health center assessment did not complete the stand, which may raise concerns about the ecological validity of the findings for the general population. Notwithstanding this caveat, it is important to acknowledge that individuals who attended the health center assessment tended to be younger and in better health compared with those who did not attend.32 This can be adduced from the fact that the mortality rate was 6.7% in the overall sample compared with 3.2% among those who completed the stand. As the derived estimates were likely to be conservative, we used survey weights in the analysis to take account of nonparticipation in the health center component.
The study also has a significant number of strengths. First, it uses a large representative sample of the community-dwelling older population and a novel measure of cardiovascular functioning (ie, active stand) that is rare in the context of an epidemiological study. Second, beat-to-beat measurement of heart rate and blood pressure affords us the resolution to explore features of the cardiovascular response to standing during a time period (ie, initial 20 s) that is of potential theoretical interest because it is arguably indexing the balance of sympathetic and parasympathetic systems. The study also benefits from the strong in-depth characterization of the sample, which means that we are able to control for a large range of variables that could potentially confound the relationship between our putative measure of cardiovascular aging and mortality.
The speed of orthostatic HRR between 10 and 20 s identifies those who are at high risk of mortality. This has important clinical applications because the speed of HRR is a clinical variable that may be useful for population screening. Although HRR is modifiable by, for example, physical activity interventions,33–36 whether modifying HRR directly reduces mortality requires further study. Future intervention studies should be designed to explore this possibility.
We thank the The Irish Longitudinal Study on Ageing (TILDA) participants who provided the data for this article.
Sources of Funding
This work was supported by a grant to the first author from the Health Research Board (HRB) of Ireland under the Interdisciplinary Capacity Enhancement (ICE/2011/7) program. Funding for the The Irish Longitudinal Study on Ageing (TILDA project is provided by the Irish Government, The Atlantic Philanthropies, and Irish Life Plc. The other authors report no conflicts.
In May 2016, the average time from submission to first decision for all original research papers submitted to Circulation Research was 14.93 days.
The online-only Data Supplement is available with this article at http://circres.ahajournals.org/lookup/suppl/doi:10.1161/CIRCRESAHA.116.308577/-/DC1.
- Nonstandard Abbreviations and Acronyms
- difference from baseline value of heart rate at time(t) where t is time in seconds after standing
- difference from baseline value of SBP at time(t) where t is time in seconds after standing
- difference from baseline value of DBP at time(t) where t is time in seconds after standing
- activities of daily living
- Cardiovascular disease
- diastolic blood pressure
- heart rate
- heart rate recovery
- speed of heart rate recovery between 10 and 20 s after standing=(ΔHR20s−ΔHR10s)
- instrumental activities of daily living
- systolic blood pressure
- Received February 19, 2016.
- Revision received June 17, 2016.
- Accepted June 21, 2016.
- © 2016 American Heart Association, Inc.
- 6.↵WHO. Collaborating Centre for Drug Statistics Methodology: Guidelines for ATC Classification and DDD Assignment 2013. Oslo, Norway: WHO; 2013.
- Tsuji H,
- Venditti FJ Jr.,
- Manders ES,
- Evans JC,
- Larson MG,
- Feldman CL,
- Levy D
- Borst C,
- Wieling W,
- van Brederode JF,
- Hond A,
- de Rijk LG,
- Dunning AJ
- Wieling W,
- van Lieshout JJ
- Finucane C,
- O’Connell MD,
- Fan CW,
- Savva GM,
- Soraghan CJ,
- Nolan H,
- Cronin H,
- Kenny RA
- Morise AP
- Dhoble A,
- Lahr BD,
- Allison TG,
- Kopecky SL
- Carnethon MR,
- Liao D,
- Evans GW,
- Cascio WE,
- Chambless LE,
- Rosamond WD,
- Heiss G
- Rowell LB
- Soraghan CJ,
- Fan CW,
- Hayakawa T,
- et al
- Cleves MA,
- Rock L
- Hlatky MA,
- Greenland P,
- Arnett DK,
- et al
- Cox DR
- Pennells L,
- Kaptoge S,
- White IR,
- Thompson SG,
- Wood AM
- Imai K,
- Sato H,
- Hori M,
- Kusuoka H,
- Ozaki H,
- Yokoyama H,
- Takeda H,
- Inoue M,
- Kamada T
- Kligfield P,
- McCormick A,
- Chai A,
- Jacobson A,
- Feuerstadt P,
- Hao SC
- Legramante JM,
- Iellamo F,
- Massaro M,
- Sacco S,
- Galante A
Novelty and Significance
What Is Known?
Impaired heart rate recovery after treadmill stress testing is an established risk factor for cardiovascular and all-cause mortality.It is hypothesized that this is because of dysregulation of autonomic balance unmasked by the test.
What New Information Does This Article Contribute?
Impaired heart rate recovery in the immediate 20 s after an orthostatic maneuver (ie, active stand from a supine position) predicts all-cause mortality independently of other established risk factors.
Heart rate monitoring (ie, with ECG) during active stand is a simpler test to determine autonomic responsiveness compared with alternative methods, such as treadmill stress testing.
Speed of heart rate recovery after physical exertion is an established risk factor for cardiovascular and all-cause mortality and is usually assessed in the clinical setting using treadmill stress testing. This study examines whether the speed of HRR in response to orthostatic challenge (ie, active stand from a supine position) has clinical value as a marker of risk. The active stand represents a potent cardiovascular stressor that elicits a rapid physiological response involving the coordinated action of several systems, including the skeletal-muscle pump and arterial and cardiopulmonary baroreflexes. Peak heart rate is reached ≈10 s after standing and declines rapidly thereafter as a result of rebounding arterial pressure. There is a particularly steep drop in heart rate between 10 and 20 s after standing and an age gradient with heart rate and blood pressure responses. We show that speed of HRR during this time window distinguishes those with existing cardiovascular disease from those who are cardiovascular disease free and has predictive power as a marker of all-cause mortality. We hypothesize that speed of HRR during this time window reflects the capability of the system to mount a vigorous response to cardiovascular challenge and to quickly re-establish homeostatic equilibrium.