OUP user menu

★ fast track ★

Impact of resting heart rate on mortality, disability and cognitive decline in patients after ischaemic stroke

Michael Böhm, Daniel Cotton, Lydia Foster, Florian Custodis, Ulrich Laufs, Ralph Sacco, Philip M. W. Bath, Salim Yusuf, Hans-Christoph Diener
DOI: http://dx.doi.org/10.1093/eurheartj/ehs250 2804-2812 First published online: 26 August 2012


Aims Recurrent stroke is a frequent and disabling event. A high heart rate is associated with cardiovascular outcomes. We investigated the effects of the resting heart rate on cardiovascular and neurological outcomes after recurrent stroke in the high-risk population of the PRoFESS study.

Methods and results A total of 20 165 patients after ischaemic stroke (mean age 66.1, SD 8.6 years) assigned to the treatment arms of the PRoFESS trial were pooled divided by quintiles of the baseline heart rate and analysed according to cardiovascular and functional outcomes after stroke: recurrent stroke and major cardiovascular outcomes such as stroke, myocardial infarction, and worsening or new-onset heart failure as well as death from cardiovascular and non-cardiovascular causes. Pre-defined endpoints were disability after a recurrent stroke, assessed with the modified Rankin scale (mRS) and the Barthel index at 3 months, and cognitive function, assessed with the Mini-Mental State Examination (MMSE) score at 4 weeks after randomization and at the penultimate visit. Patients in the two highest quintiles of heart rate (77–82 and >82 b.p.m.) were at a higher risk for total death [hazard ratio (HR) 1.42, 95% CI 1.19–1.69 and HR 1.74, 95% CI 1.48–2.06, P < 0.0001] compared with the lowest quintile. Similar results were observed for vascular death [71–≤76 b.p.m., HR 1.39 (1.11–1.74), P < 0.0001] and non-vascular death [from >82 b.p.m., HR 1.66 (1.29–2.13), P = 0.0016]. Myocardial infarction (P = 0.7084) and recurrent stroke (P = 0.1379) were not significantly associated with the baseline heart rate. Hazard ratios were adjusted to multiple confounders including the baseline blood pressure. In the group of patients with a recurrent stroke, an association of a lower heart rate to better outcomes was measured with the Barthel index across all heart rate groups. In addition, there was a significant association of the baseline heart rate to the occurrence of significant cognitive decline according to an MMSE score ≤24 points at 1 month and at the penultimate visit or a decline of ≥2 points between these two time periods. Better independence score at a low heart rate were observed.

Conclusion The heart rate is a risk indicator for mortality in patients with stroke and, importantly, a low heart rate is associated with a better functional outcome and less cognitive decline after an ischaemic stroke.

Trial registration: ClinicalTrials.gov, number NTC00153062.

  • Heart rate
  • Stroke
  • Cardiovascular outcomes
  • Cognitive decline
  • Dementia

See page 2761 for the editorial comment on this article (doi:10.1093/eurheartj/ehs260)


Several epidemiological and clinical studies indicate that the intrinsic heart rate at rest predicts cardiovascular death in the general population,1 and in patients with hypertension,2 coronary artery disease,3 high-cardiovascular risk,4 and heart failure.5 Experimental studies have shown that heart rate reduction with the If-channel inhibitor, ivabradine, slows atherosclerosis6 and reduces the stroke size.7 Although the effects of blood pressure on vascular outcomes, mortality, and functional outcomes after stroke are well characterized,810 no such information is available for the heart rate.

The Prevention Regimen for Effectively Avoiding Second Stroke trial (PRoFESS) compared the effectiveness of prophylactic treatment with aspirin and extended release dipyridamole vs. treatment with clopidogrel as well as treatment with telmisartan or placebo in a two by two factorial design in patients who have had an ischaemic stroke.11 There was no evidence that telmisartan was superior to placebo12 or that the combination of aspirin and extended release dipyridamole was superior to clopidogrel13 in the prevention of recurrent stroke or with regard to neurological outcomes.14 This post hoc analysis of PRoFESS aimed at evaluating associations of the resting heart rate at baseline with cardiovascular outcomes and neurological outcomes among patients who recently experienced an ischaemic stroke or suffered a recurrent stroke. Data of patients from the treatment arms, which did not show different outcomes, were pooled for this analysis.


In brief, from September 2003 to July 2006, 20 332 patients from 695 centres in 35 countries were assigned to the different treatment arms after they had had a non-cardioembolic ischaemic stroke. Patients who experienced an ischaemic stroke within 120 days prior to randomization were aged 55 years or older or patients aged 50–54 years had to have two additional cardiovascular risk factors. The trial had a two by two factorial design to compare for treatment regimens containing extended release dipyridamole plus aspirin compared with clopidogrel or telmisartan, compared with placebo. All patients were on optimum medical treatment at the discretion of the investigators including drugs for controlling blood pressure. Ischaemic stroke qualified for inclusion, if there was a focal neurological deficit of cardiovascular origin >24 h. In patients whose symptoms persisted <24 h, they had to have evidence of a current ischaemic stroke on computed tomography or magnetic resonance imaging. Primary haemorrhagic stroke, severe disability after the qualifying stroke, or contraindications to one of the drug treatments or other factors, as summarized in the design paper,11 were reasons for the exclusion of the patients. After qualifying, stroke patients could be randomized, being still in the hospital, up to the follow-up of 120 days after stroke. The following evaluations were done at 1 week, and at 1, 3, and 6 months as well as every 6 months thereafter. Patients who were unable to come to the hospital were contacted by telephone. The heart rate was taken at entry in a sitting position together with blood pressure. Blood pressure was measured by using a standard and validated Omron sphygmomanometer (Omron Health Care Inc.) with an appropriately sized cuff. Baseline data were to be collected the same day as randomization. The median time from the index stroke to the baseline heart rate assessment was 15 days, 7 and 39 days being the 25th and 75th percentiles. The blood pressure and the heart rate was measured twice at least 2min apart and averaged.11 At the follow-up visits, vital signs were registered, but in a less standardized way. The median duration of the follow-up was 2.4 years, 2.0 and 3.0 years being the 25th and 75th percentiles.

The primary outcome was a recurrent stroke of any type. The composite of stroke, myocardial infarction, and death from vascular causes were determined as secondary outcomes. An adjudication committee blinded to treatments adjudicated the primary and the secondary outcomes on a time to first event basis.

The functional outcome after the recurrent stroke was evaluated with the modified Rankin scale (mRS) 3 months after a recurrent stroke. The Barthel index was monitored 3 months after the recurrent stroke. The mRS is a disabling scale ranging from 0 (no disability) to 5 (patients are bedbound requiring constant care) or 6 (died). The Barthel index measures the activity of daily living with scores ranging from 0 (complete dependence) to 100 (independence). The cognitive function was evaluated in all patients with a Mini-Mental State Examination (MMSE). The score ranged from 0 to 30, with lower scores indicative of a greater degree of cognitive impairment with 27.6 points as evidence for being cognitively intact and a score of 24 points or lower indicative of some degrees of cognitive impairment. The MMSE was done 1 month after randomization, at 2 years, and at the penultimate visit. In PRoFESS, the different treatment arms did not show any difference on cardiovascular outcomes or on disability and cognitive function. To define the role of the resting heart rate in cardiovascular and neurological outcomes, data of all enrolled patients were combined and included in these analyses following the intention to treat principle.

This study evaluated outcomes according to quintiles of the heart rate at baseline. The primary outcome was a recurrent stroke of any type and the secondary endpoints of major cardiovascular outcomes (stroke, myocardial infarction, death from cardiovascular causes) were evaluated on the time to first event basis. Neurological function and cognitive outcomes were evaluated as described previously.

Statistical analysis

Baseline and demographic and clinical covariates were pre-selected based on prior studies of factors that influenced the occurrence of vascular events and outcomes after the ischaemic stroke and are presented by heart rate quintiles at baseline, with means and SDs for continuous variables and percentages for categorical variables. χ2 tests were used for categorized variables and the Kruskal–Wallis test was applied for continuous variables. Neurological outcomes are presented similarly. Hazard ratios (HRs) were also calculated. Individuals in the lowest quintile (at a heart rate ≤64b.p.m.) were used as the reference group. For neurological outcomes, the Kruskal–Wallis test was used for continuous variables and χ2 test for categorized variables. ANOVA was used for the modified ranking scale score. Covariate selection was performed by including the following pre-specified covariates into an initial stepwise model: blood pressure, age, sex, history of prior stroke (before qualifying stroke), history of congestive heart failure, history of diabetes, history of MI, and other covariates presented in Table 1. A backward elimination method selected covariates with a P-value level of <0.10 to enter into the model and a P-value level of <0.10 to remain with the heart rate in the final model for analysis. The assumption of proportional hazards was tested for the death endpoint by examining the baseline heart rate by the logarithm of time interaction. Each of the indicators for the upper four quintiles for the baseline heart rate was included in the model. The overall interaction test yielded P = 0.09. On an individual covariate basis, the interaction for the baseline heart rate >82 b.p.m. by log (time) was a concern (P = 0.02). So, a Kolmogorov-type supremum test was performed for each of the covariates and that yielded P = 0.06 for the baseline heart rate >82 b.p.m. covariate. In addition, an inspection of the log{−log[S(t)]} plot vs. log(time) was performed and the conclusion was that the assumption of proportional hazards was met.

View this table:
Table 1

Baseline characteristics of study participants by quintiles of the resting heart rate

All endpoints reported were pre-specified in a publication statistical analysis plan that was written after study results were reported, except the statistical test for the 3-month post-stroke mRS, which was requested after initial results were reviewed. The overall significance level for the study was 0.05 using the two-sided test. All analyses were performed using the SAS statistical software version 9.2 (SAS Institute Inc.).

Role of funding

The sponsor was involved in the study design, collection, and interpretation of the data. The sponsor was not involved in the decision to submit the paper for publication. All authors had full access to the data and analyses and vouch for the currency and completeness of the data reported. All authors were involved in the final decision to submit the manuscript.


All PRoFESS patients with the baseline heart rate collected (20 165 out of the 20 332 patients randomized) entered the analysis. The mean age was 66 (SD 8.6) years and 36% of the population were women. The baseline characteristics including ethnic groups and pre-existing diseases according to quintiles of the heart rate at baseline were summarized in Table 1. Several of these baseline parameters were similar. However, patients with a high heart rate were younger and more likely women and less likely to drink alcohol or to smoke. Beta-blocker treatment was more prevalent in low than in high heart rate groups. Interestingly, patients with a higher heart rate have a higher TOAST classification indicative of large artery atherosclerosis as well as the difference according to the mRS score at baseline and the NIHSS score (Table 1).

When the total population was divided by heart rate quintiles at baseline, the primary endpoint of recurrent stroke was not significantly different between all heart rate groups (log rank P = 0.1379). This holds true for the unadjusted data as well as for data adjusted for age, sex, tobacco use, exercise, small artery occlusion, TOAST classification, mRS, NIHSS and previous medical conditions, use of medications, and blood pressure (Table 2, Figure 1A). There was also no significant association with stroke, myocardial infarction or new onset or worsening of heart failure (Table 2). Total death was associated with the resting heart rate (Figure 2A). The threshold of an increased risk for adjusted and unadjusted death occurred at a heart rate of 71–≤ 76 b.p.m. with an adjusted HR of 1.32 (95% CI 1.11–1.56). At a heart rate >82 b.p.m., the risk of death increased to an HR of 1.74 (95% CI 1.48–2.06) after the adjustment for covariates. Similar data were obtained for cardiovascular death (Figure 2B, Table 2). The HR was 1.39 (1.11–1.74) at a resting heart rate of 71–≤76 b.p.m., and amounted to 1.78 (1.44–2.22) at a heart rate >82 b.p.m. (Table 2). A similar but a weaker association was observed for non-cardiovascular death with a significant HR of 1.66 (1.29–2.13) at a heart rate >82 b.p.m. (Figure 2C, Table 2). To study the interaction of systolic or diastolic blood pressure with the heart rate, the interaction of blood pressure with the heart rate was examined. Since the correlation was low, blood pressure was included in the covariates for selection into the adjusted model. Adding systolic or diastolic blood pressure, there was no change of risk, indicating that the effects of the heart rate on risk are independent of the blood pressure (not shown).

View this table:
Table 2

Hazard ratios of a primary outcome stroke and the secondary outcomes of myocardial infarction, new-onset or worsening of heart failure as well as death, cardiovascular death, and non-cardiovascular death

Figure 1

Kaplan–Meier curves for cumulative probability of recurrent stroke (A) or myocardial infarction (B), according to the groups defined by quintiles by the resting heart rate. Q1–5 denote the quintiles of the baseline heart rate. Hazard ratios were calculated with the use of the Cox model, which was adjusted to baseline characteristics.

Figure 2

Kaplan–Meier curves for the cumulative probability of death (A), cardiovascular death (B), and non-cardiovascular death (C). Q1–5 denote the quintiles of the baseline heart rate. Hazard ratios were calculated with the use of the Cox model, which was adjusted to baseline characteristics.

To determine whether the baseline heart rate has an impact on the patients’ global disability scale according to the mRS, individuals with a recurrent stroke (n = 1627) were scaled into seven categories. There was a better functional outcome 3 months after the recurrent stroke with a lower heart rate (P = 0.0191, Table 3, Figure 3). Furthermore, quintiles of the heart rate were associated with disability according to the mRS at baseline (P < 0.001) and 3 months after the recurrent stroke (P = 0.0002, Figure 3). To measure an index of daily life activities and independence, the Barthel index was evaluated 3 months after the first stroke. We could show that after the index stroke in 1529 patients, the Barthel index was higher in patients in the first lower quintiles of the heart rate (P = 0.0002, Table 3).

View this table:
Table 3

Neurological outcomes by baseline heart rate quintiles

Figure 3

Comparison of modified Rankin scale scores at baseline and at 3 months after the recurrent stroke according to the baseline heart rate and quintiles (Q1–5); P-values were calculated from a one-way ANOVA test.

Cognitive function was evaluated with an MMSE at 1 month and after the penultimate visit. It is shown that heart rate quintiles were related to impaired cognitive function (MMSE ≤24) at 1 month and during the follow-up until the penultimate visit (P = 0.0001). Furthermore, at a heart rate >82 b.p.m. more patients had a decrease of two points in the MMSE between 1 month and the penultimate visit (Table 3). The judged cut-off value is ∼70 b.p.m. for cognitive decline according to the MMSE.


Our results show that in patients after a first stroke, a higher heart rate at baseline is associated with an increased risk of total death, vascular death, and non-vascular death starting at a rather low heart rate threshold of 76 b.p.m., while there was no significant association with recurrent stroke, myocardial infarction, and new-onset or worsening congestive heart failure. Interestingly, poor functional independence in patients with a recurrent stroke according to the Barthel index score and cognitive decline associated with a cut-off of the MMSE ≤24 were significantly associated with an increasing resting heart rate and a trend towards poorer functional outcome of patients after the first stroke according to the mRS.

Stroke is a disabling condition regarded as the second most frequent cause of death leading to disability, cognitive impairment, and a huge economic burden on health care systems.8 Hypertension is regarded as one leading risk factor for stroke8 and reducing blood pressure has been shown to reduce morbidity and mortality due to stroke in large hypertension trials.9,15 Recent studies have associated increases in blood pressure but also resting heart rate and changes thereof at the follow-up in response to stress as risk indicators for cardiovascular outcomes.1620 The heart rate is regarded as a risk indicator in hypertension2 and cardiovascular disease4 and as a modifiable risk factor in heart failure5 as heart rate lowering with an If-channel inhibitor reduces events.21 In an experimental model subjected to mental stress, there was an impairment of endothelial function with an increase in the experimental stroke size but, in turn, a reduction in the stroke size after heart rate lowering.7 The high-risk cardiovascular population of the PRoFESS trial appeared appropriate to investigate the effect of heart rate as a predictor of cardiovascular outcomes and also on functional neurological outcomes after an ischaemic recurrent stroke.

Herein, we report that vascular death and also non-vascular death adjusted to multiple confounders including blood pressure are associated with the resting heart rate. These data show that in elderly patients after stroke, the heart rate is a general predictor of mortality, beyond specific effects in well-characterized pathologies in vascular disease.22,23 No significant effects could be found for recurrent stroke and myocardial infarction. One study showed that the baseline heart rate >70 b.p.m. was associated with plaque rupture in coronary arteries.24 Furthermore, the BEAUTIFUL study has shown an association of a heart rate >70 b.p.m. to non-fatal myocardial infarction in a population after a first myocardial infarction with impaired left ventricular function.3 This population might have been more sensitive to a second myocardial infarction and the heart rate in the overall population of PRoFESS might have been too low to show associations with myocardial infarction in this post-stroke population. This might explain why this analysis does not show heart rate associations with myocardial infarction. However, cardiovascular death and total death might be carried by multiple micro- and macro-vascular pathophysiologies. The heart rate is also a marker of early ageing and death in many animal species and potentially representing a marker of general ageing,25 potentially modifiable by pharmacological heart rate reduction.26

In mice, a fixed occlusion of the cerebral arteries is a model for the stroke size, in which different stress models were leading to increases in stroke sizes, which could be reduced by heart rate lowering associated with an improvement of endothelial function.7 Therefore, not only plaque rupture, but also the general vascular endothelium might be important.27 Interestingly, collateral growth after hindlimb occlusion was enhanced by a lowering heart rate due to a reduced expression of anti-angiogenetic cytokines and pro-inflammatory cytokines.28 Therefore, vascular compensatory mechanisms might be induced at a lower heart rate. Therefore, we set out to investigate functional neurological outcomes after stroke, which could be the clinical consequence of stroke size reduction as shown in experimental studies. Indeed, the present analysis showed that there was a significant association with functional independence determined by the Barthel index score according to quintiles of the heart rate. Furthermore, there was a significant association between the heart rate and the modified Rankin scores 3 months after the recurrent stroke. Fewer patients with severe cognitive decline (>2 points decline in the MMSE) at 1 month between the index stroke and at the end of the study visit in the lower heart rate quintile were identified. Consistently, less number of patients declined to an MMSE ≤24 points indicative of dementia. These results show an improvement of functional consequences after stroke or lower stroke sizes despite the lack of prevention of recurrent stroke by a low heart rate. An increased heart rate in a model with intermittent stress developed intracerebral oxidative stress29 indicated by an up-regulation of brain lipid hydroperoxides, superoxide production, and increase in the angiotensin II-type AT1 receptors.7,29 Heart rate lowering improved brain capillary density as judged by CD31-expressing cells in the brain of stressed mice.7 Although these mechanisms cannot be directly shown in patients, these studies might provide the mechanistic explanation for the functional improvement of neurological outcomes after stroke according to the heart rate.

Our analysis from the PRoFESS trial has limitations. Caution is advised, because this analysis is a retrospective post hoc analysis of a randomized trial with a neutral outcome and patients were not assigned in a randomized fashion to different groups of heart rate. However, differences in baseline values are small and we appropriately took care of this by adjustment methods. Furthermore, the size of the trial allowed adequate power compared with previous analyses to detect the association of the heart rate to outcomes. The association with outcomes after stroke were observed in heart rates <60 to >83 b.p.m., which are quite low and have to be regarded as being in the normal range. This analysis provides the first evidence that different physiological heart rates are risk indicators for cognitive decline and a general mortality marker in patients after an ischaemic stroke independent from the blood pressure. One might suggest that in the trial heart rate and changes of the heart rate at the follow-up vs. the heart rate at baseline might be more closely related to outcomes. However, the heart rate at follow-up visits was taken in a less standardized fashion. Therefore, we cannot provide reliable information of heart rate changes. Furthermore, heart rate changes might be dependent on developing comorbidities during this long-term follow-up. Therefore, we took the conservative approach to explore the resting heart rate close to the index stroke as a risk predictor for long-term outcomes.

It is concluded that a baseline heart rate >76 b.p.m. is associated with mortality in the population of patients after the ischaemic stroke. Myocardial infarction and recurrent stroke are not associated with these heart rate levels, but functional neurological outcomes after the recurrent stroke are better at a low heart rate and could be an indicator of smaller stroke sizes as suggested by experimental studies. This analysis is hypothesis generating and sets the stage for the evaluation of potential pharmacological interventions to reduce the heart rate in patients after an ischaemic stroke.


PROFESS was funded by Boehringer Ingelheim, Germany.

Conflict of interest: M.B. has received study support and honoraria from: Astra Zeneca, Bayer AG, Boehringer Ingelheim, Novartis, Pfizer, Sanofi-Aventis, Servier, Adrian-Medtronic, Daiichi-Sankyo, MSD, AWD Dresden, and Berlin-Chemie. P.M.W.B. has received study support and honoraria from Boehringer Ingelheim, Lundbeck, Mitsubishi, Phagenesi, and ReNeuron.


View Abstract