Introduction

Resting heart rate (RHR) is a determinant of cardiovascular function and contributes to the pathogenesis of myocardial and vascular disease [21]. Increased heart rate is closely linked to endothelial dysfunction and atherosclerosis in experimental animal models and in humans [6, 9].

Epidemiological and interventional studies investigated the association between RHR and outcomes in different populations, and there is evidence from large studies that elevated RHR is predictive of cardiovascular outcomes, independently of standard risk factors or co-morbidities, in patients with established cardiovascular diseases [4, 10, 18, 22, 25].

Most epidemiological studies addressing RHR in a primary prevention setting were conducted prior to the broad implementation of cardiovascular protective therapies [10, 18, 22, 25]. Data from contemporary cohorts are sparse [26]. In patients with heart failure, RHR has emerged both as a powerful marker of risk in patients with impaired ejection fraction and a treatment target that is easily available through clinical examination, thus qualifying heart rate as being a risk factor [2–4, 35]. Therefore, we evaluated the association of RHR with all-cause mortality and coronary events in the Heinz Nixdorf risk factors, evaluation of coronary calcium, and lifestyle (RECALL) study, a population-based, prospective cohort study of the comparative value of risk stratification techniques for cardiovascular events [34].

Methods

Participants

Participants were randomly selected from mandatory city registries in Essen, Bochum, and Mülheim and invited to participate in the study as previously reported [11, 12]. Physician- or self-referral was not allowed to avoid selection bias. A total of 4814 subjects aged 45–75 years (50 % females) were included between December 2000 and August 2003. All subjects with physician-diagnosed coronary artery disease (CAD), i.e. a history of myocardial infarction or coronary revascularization (n = 327) or stroke (n = 135) were excluded from the study as well as persons with atrial fibrillation in the resting ECG (n = 80). In six individuals a resting ECG was unavailable, and 227 persons had no valid information on heart rate lowering medication, leaving 4091 persons for the present analysis. All participants provided written informed consent, and the study was approved by the Ethical Committee at the University Essen, Germany.

Heart Rate, Cardiovascular Risk Factors and Questionnaires

RHR was obtained from standardized automated measurements from 12-lead resting ECGs after 5 min rest using a MAC 5000® ECG recorder (GE Healthcare, Freiburg, Germany) as previously reported [29]. Blood pressure was determined from the mean of the second and third measurements taken at least 3 min apart (Omron 705-CP, OMRON, Germany). Body mass index (BMI) was calculated from standardized measurements of height and weight. Current smoking was defined as a history of cigarette smoking during the past year. Former smoking was defined as a history of ever smoking. Standard enzymatic methods were used to measure total cholesterol, LDL- and HDL-cholesterol, and triglycerides. Participants were considered diabetic if they reported a physician diagnosis of diabetes or were taking anti-diabetic medication. HsCRP was measured using a standardized assay (Roche Diagnostics, Basel, Switzerland).

All subjects were queried about known cardiovascular disease (CVD) using a physician-based questionnaire and about regular cardiovascular medication including anti-hypertensive medication, heart rate lowering medication including beta-blockers, glycosides, and dihydropyridine-calcium channel blockers.

Electron-beam computed tomography (EBCT)

To quantify CAC, non-enhanced EBCT scans were performed with a C 100 or C-150 scanner (GE Imatron, South San Francisco, US). Prospective ECG-triggering was done at 80 % of the RR-interval. Contiguous 3 mm thick slices to the apex of the heart were obtained at an image acquisition time of 100 ms. CAC was defined as a focus of at least four contiguous pixels with a CT density ≥130 Hounsfield Units. The CAC Agatston Score was computed by summing the CAC scores of all foci in the epicardial coronary system and was log-transferred because of skewed data distribution [1, 27, 33]. The CAC score was not communicated to either the participants or their treating physicians.

Follow-up

Annual postal questionnaires assessed the morbidity status during follow-up, i.e. medication, hospital admissions, and outpatient diagnoses of CVD. Self-reported incident cardiovascular morbidity and fatal events were validated by review of hospital records and records of the attending physicians (see below). All death certificates of the three cities under study were regularly screened. In parallel, deceased participants were tracked back to obtain as much information as possible to verify the causes of death.

Study endpoints and verification of study endpoints

Primary endpoints for this study were based on unequivocally documented incident coronary events that met predefined study criteria [11, 12, 34]. We considered a myocardial infarction event based on symptoms, electrocardiographic signs, and enzymes [levels of creatine kinase (CK-MB)] as well as troponin T or I, and necropsy as: (1) non-fatal acute myocardial infarction or (2) coronary death [11, 12]. For all primary study endpoints, hospital and nursing home records including electrocardiograms, laboratory values, and pathology reports were collected. For deceased subjects, death certificates were collected, and interviews with general practitioners, relatives, and eyewitnesses were undertaken where possible. Medical records were obtained for all reported endpoints. An external criteria and endpoint committee blinded for conventional risk factor status and CAC scores reviewed all documents and classified the endpoints thereafter.

Statistical analysis

Demographic data and risk factors are expressed as means (standard deviation, SD) or medians (25th, 75th percentile), frequencies are given as counts (%). Differences in proportions were statistically evaluated using Chi square or Fisher’s exact test, trends in proportions using the Cochran-Armitage trend test; location measures of continuous quantities were compared using Student’s t test or Mann–Whitney U statistics.

Observed 12-year cumulative risks are given in quartiles of heart rate (Q1–4) and in 10 bpm categories of heart rate, plotting observed risks, and a parabolic fit curve to assess characteristic deviations from linearity. We used univariate and multivariable Cox proportional hazard regression to calculate unadjusted and adjusted hazard ratios (HRs) and corresponding 95 % confidence intervals (CI) for the occurrence of endpoints (death or coronary event). HR estimates for heart rates were adjusted for conventional risk factors, i.e. age, sex, hypertension, current smoking, diabetes, BMI, known PAD, LDL-cholesterol, lipid-lowering medication, heart rate lowering medication (where applicable), and CAC. Kaplan–Meier survival curves were calculated per quartile of heart rate, and statistically evaluated using a Log-rank test of trend. All calculations were performed with SAS (Cary, NC, version 9.4).

Results

Study population

Baseline characteristics and distribution of CVD risk factors by RHR quartile are shown in Table 1. Mean RHR in the study population was 69.7 ± 11 bpm. At study entry, 18.2 % of subjects were taking heart rate lowering medication. Individuals with lower RHRs were more likely to be on heart rate lowering drugs. In the lowest quartile (Q1) there were more men than in Q2–4. The prevalence of cardiovascular risk factors, e.g. systolic blood pressure, LDL-cholesterol, diabetes mellitus, and hsCRP increased with heart rate quartiles (Table 1). Q4 contained almost twice as many individuals with diabetes mellitus than Q1 and hsCRP levels were almost twice as high. CAC scores and the prevalence of PAD showed a J-shaped distribution across heart rate quartiles since persons with heart rate lowering medication had a higher prevalence of CAC and known PAD than those without (77 versus 66 %, p < 0.0001, and 2.5 versus 1.3 %, p = 0.03, respectively). Persons with heart rate lowering medication had lower heart rates than those without (66.0 ± 11 versus 70.6 ± 11 bpm, p < 0.001).

Table 1 Baseline characteristics and distribution of CVD risk factors by heart rate quartiles
Full size table

Predictors of events

During a median follow-up of 10.5 years (Q1/Q3: 10.2/13.0 years), 159 (3.9 %) subjects developed a coronary event, and 398 (9.7 %) subjects died of any cause. Persons without any events had a similar mean heart rate (69.5 ± 11 bpm) as persons with coronary events (69.9 ± 11 bpm, p = 0.51) but lower mean heart rates than persons who died (72.3 ± 13 bpm, p < 0.0001). In the entire cohort using a multivariate full model, age, current smoking, known PAD, log(CAC + 1), and RHR [Q4 versus Q1, HR = 1.64 (1.22–2.22), see Table 2] were independent predictors of all-cause mortality (hazard ratios other than for heart rate not shown). Age, sex, known PAD, and log(CAC + 1), but not RHR, were independent predictors of a coronary event (hazard ratios not shown).

Table 2 Incidence of events and hazard ratios by heart rate quartile in the entire cohort
Full size table

Persons with HR lowering medication had higher all-cause mortality rates (Fig. 1a; Table 3) and higher coronary event rates (Fig. 1b; Table 4) than persons without. This was also seen in HR quartiles for all-cause mortality except for Q4 and for coronary events in Q1, but not in Q2–4 (data not shown).

Fig. 1
figure 1

Kaplan–Meier time-to-event plots for all-cause mortality (a) and coronary events (b) in the subgroups of persons with heart rate lowering medication and persons without heart rate lowering medication

Full size image
Table 3 Incidence of events and hazard ratios by heart rate quartile in individuals with heart rate lowering medication
Full size table
Table 4 Incidence of events and hazard ratios by heart rate quartile in individuals without heart rate lowering medication
Full size table

In individuals with heart rate lowering medication, only age and log(CAC + 1) were independent predictors of all-cause mortality, with known PAD having borderline statistical significance [HR = 2.39 (0.97–5.92), p = 0.059]. Independent predictors of coronary events were sex and log(CAC + 1), while a high hazard in persons with known PAD [HR = 3.03 (0.95–9.68), p = 0.061] did not reach statistical significance.

In persons without heart rate lowering medication, age, smoking, diabetes, log(CAC + 1), and RHR [HR = 1.84 (1.28–2.65), p < 0.0001 for Q4 versus Q1] were independent predictors of all-cause mortality. In this subgroup, independent predictors of coronary events were age, sex, smoking, known PAD, and log(CAC + 1), but not RHR.

Effect of heart rate quartiles on outcome

In the entire cohort, all-cause mortality risk but not coronary event risk increased with heart rate quartiles (Table 2; Fig. 2a, d). The increased mortality risk was particularly driven by events in persons without heart rate lowering medication (Table 4; Fig. 2c), but not in persons with heart rate lowering medication (Table 3; Fig. 2b, e).

Fig. 2
figure 2

Kaplan–Meier time-to-event plots for all-cause mortality (a–c) and coronary events (d–f) in the prespecified subgroups (a, d entire cohort, b, e persons with heart rate lowering medication, c, f persons without heart rate lowering medication)

Full size image

Risk associated with heart rate ≥70 vs <70 bpm

In the entire cohort, individuals with heart rate ≥70 bpm had 50 % increased hazards for dying than persons with heart rate <70 bpm (HR = 1.53, 1.23–1.90, p = 0.0001). The HR for coronary events was 1.07 (0.77–1.51), p = 0.68, in persons with heart rate ≥70 versus <70. In persons with heart rate lowering medication, the risk of all-cause mortality was HR = 1.25 (0.81–1.95), p = 0.32, and of coronary events HR = 0.76 (0.35–1.67), p = 0.49, for heart rate ≥70 versus <70 bpm, respectively. In persons without heart rate lowering medication, the risk of all-cause mortality was HR = 1.68 (1.30–2.18), p < 0.0001, and of coronary events HR = 1.20 (0.82–1.77), p = 0.35, for heart rate ≥70 versus <70 bpm, respectively.

Risk per 5 and 10 bpm increase in resting heart rate

For the entire cohort, an increase of heart rate by 5 bpm was associated with increased all-cause mortality in unadjusted, age, and gender adjusted as well as traditional risk factor adjusted models [HR 1.10 (1.05–1.15), p < 0.0001, full model]. The risk for coronary events was not significantly increased [HR 1.01 (0.94–1.09), p = 0.82, full model]. In persons with heart rate lowering medication, the HR of all-cause mortality was 1.02 (0.92–1.13), p = 0.70, and for coronary events HR was 0.97 (0.82–1.14), p = 0.71 for an increase in heart rate by 5 bpm. In persons without heart rate lowering medication, the HR of all-cause mortality was 1.13 (1.07–1.20), p < 0.0001, and for coronary events HR was 1.02 (0.94–1.11), p = 0.60 for an increase in heart rate by 5 bpm.

To better understand the relationship between RHR and outcomes across the RHR spectrum, we also plotted events in 10 bpm RHR intervals (Fig. 3a–f). In the entire cohort—driven by persons without heart rate lowering medication—the relationship between RHR and all-cause mortality was non-linear, while we observed no association of RHR with coronary events. These results were confirmed when repeating the analyses for either sex.

Fig. 3
figure 3

Relationship between heart rate as a continuous variable, all-cause mortality (a–c), and coronary events (d–f) as a result of modeling with parabolic fits in the prespecified subgroups [a, d entire cohort, b, e persons with heart rate lowering medication, c, f persons without heart rate lowering medication; incidence of events in heart rate category (10 bpm intervals), parabolic fit (95 % CLM)]

Full size image

Discussion

The main finding of the present study is a positive and continuous association between RHR and mortality but not coronary events in a contemporary population-based prospective cohort. The study demonstrates that risk clearly increases at heart rates far below the common definition of tachycardia. These results support earlier studies that showed an association between RHR and mortality endpoints in cohorts free of CAD and were confirmed when repeating the analyses for either sex [5, 24]. Importantly, the observed association between baseline heart rate and risk is continuous and displays a J-shaped relation with a signal of increasing risk for heart rate below 60 bpm. Similar observations were found in other study populations indicating a non-linear relation between heart rate and mortality risk [17]. An increased risk in the lower heart rate range may reflect an indicator of underlying cardiovascular disease and may partially be associated with bradycardia related adverse events. Even though these findings represent epidemiologic data, the observation that risk may increase with low heart rate is supported by data from interventional studies. Recently, an interventional trial testing the prognostic importance of a pronounced heart rate reduction induced by ivabradine in a CHD population showed an increase in the incidence of the primary endpoint among patients with angina [14].

Early epidemiologic studies such as NHANES I or the Framingham study suggested an association between RHR and risk for coronary mortality [19, 25]. In our analysis RHR did not affect the risk for coronary events, a finding that is supported by contemporary studies investigating populations without underlying cardiovascular disease [13]. Data from the National FINRISK Study show a significant association between heart rate and risk associated with coronary heart disease. However, a fundamental reduction in the strength of the relationship was seen when myocardial infarction was integrated in the analysis [5]. Interestingly even in high risk populations with clinically manifest atherosclerotic vascular disease, RHR was associated with increased risk for mortality but not for myocardial infarction or stroke [23, 26]. Furthermore, a targeted heart rate reduction did not affect coronary events and outcome in patients with stable coronary artery disease in an interventional trial [14]. Differing findings regarding coronary events may result from methodical differences concerning endpoint definition. To our knowledge, the Heinz Nixdorf RECALL study is the first and only study that included enzymatic methods [levels of creatine kinase (CK-MB), troponin T or I] or characteristic signs in resting 12-lead ECG to define a myocardial infarction event. All comparable studies used ICD codes based on national registries or death certificates resulting in attenuated precision.

A considerable proportion of participants received heart rate lowering medication, which constitutes a fundamental part of several CV disease drug regimens. Consequently, we stratified risk analysis for individuals with and without such medication. Interestingly, the observed effect of heart rate on mortality risk was essentially driven by persons not taking heart rate lowering medication. Moreover, overall mortality risk and coronary event risk was higher in persons taking heart rate lowering medication. Our data suggest that persons with underlying CV disease—other than excluded known myocardial infarction, coronary revascularization, stroke, or atrial fibrillation—were the ones receiving heart rate lowering medication. It appears that such medication may attenuate risk, but not enough to achieve risk levels of otherwise healthy individuals. Hence, the effect of heart rate on mortality risk can especially be seen in low risk persons where risk is not predominantly driven by risk factors with a higher impact than that of heart rate. To the best of our knowledge, the present analysis is the first that analyzed risk in relation to heart rate lowering medication in the general population. Findings from comparable studies would therefore be of great interest.

Despite a wealth of data from epidemiological and interventional studies regarding the relation between RHR and cardiovascular risk, a distinct heart rate-threshold at which risk increases and the quantitative relation between heart rate increase and outcome has been debated. In patients with coronary artery disease and impaired left-ventricular function, elevated RHR of 70 bpm or greater increases cardiovascular mortality and the risk for coronary events [16]. In the Heinz Nixdorf RECALL study population, individuals with heart rates of 70 bpm or greater had an increased risk of all-cause mortality. Again, this risk was more pronounced in individuals without heart rate lowering medication. Whether RHR acts as a marker of underlying disease or may be seen as a risk factor independent of established cardiovascular risk factors has been a matter of debate. Here, our data show a relevant association between RHR and all-cause mortality after adjustment for common risk factors. Heart rate remained a prognosticator even after adjustment for atherosclerotic manifestations such as PAD and CAC.

In the HNR study population, increased RHR was associated with the presence of cardiovascular risk factors. Individuals with a high heart rate were more likely affected by hypertension and diabetes mellitus and showed higher LDL-C and triglyceride levels. Atherosclerotic manifestations (CAC and PAD) and higher CRP levels were more prevalent in high than in low heart rate groups. These relationships are consistent with previous studies that showed positive correlations between an unfavorable pattern of risk factors and heart rate [36]. The underlying mechanisms of such associations are not fully understood and conclusions about cause and effect remained speculative. Hypertension and insulin resistance may contribute to an autonomic imbalance (e.g. sympathetic overactivity) that causes increased RHR [30]. On the other hand, animal models such as stress-induced increase of heart rate suggest that cardiovascular structural alterations can be the symptom of increased heart rate [8]. Mechanistically, several potential links between RHR and cardiovascular mortality have been proposed. One of the most relevant seems to be an interference of increased heart rate with vascular integrity as experimental and clinical data suggest that sustained elevation of heart rate may contribute to the pathogenesis of vascular disease [9]. In animal studies, accelerated heart rate was associated with vascular oxidative stress and inflammation, endothelial dysfunction, acceleration of atherogenesis, and vascular stiffness [6, 7, 32]. In humans, high RHR was shown to be closely linked to systemic inflammation, arterial stiffness, and markers of endothelial dysfunction. Increased heart rate predisposed to coronary plaque rupture [20, 31, 37]. Earlier analyses of the present cohort characterized RHR as an independent predictor of coronary artery calcium (CAC), a measure of coronary plaque burden [28]. However, although there is ample experimental evidence for direct vascular effects, the potential weight for clinical practice remains elusive since interventional clinical studies investigating effects of a targeted heart rate reduction on vascular endpoints are lacking to date. Even with regard to large randomized controlled trials which assessed the prognostic role of heart rate reduction in patients with established coronary heart disease transformation of experimental findings into clinical impact still remains a future target [14, 15].

A possible limitation of the present study is the restricted number of HR recordings that was limited to a single measurement. As heart rate is susceptible to a large number of influences and effectors multiple recordings collected at shorter intervals could have shown a longitudinal course of RHR and may have increased the association between RHR and risk for events.

Conclusion

In this prospective study of an unselected general population without known CAD, elevated heart rate is an independent marker of subsequent death. The relationship between RHR and risk was graded and persisted after adjustment for risk factors, PAD, and CAC. The association was particularly pronounced in individuals without heart rate lowering medication. RHR is easily and widely available and can therefore be used as a prognostic marker in the general population. Interventional trials are warranted to test whether heart rate can be used not only as a powerful marker of risk but also as a treatment target in populations with and without known cardiovascular disease.