Mortality in the entire sample and among participants taking or not taking drugs affecting heart rate. Light bars indicate rate among those not taking HR-modifying drugs; dark bars indicate rate for participants taking drugs affecting HR. Error bars indicate 95% CIs.
eFigure. Incidence of cardiovascular disease by heart rate category, per 1000 person-years
Dharod A, Soliman EZ, Dawood F, Chen H, Shea S, Nazarian S, Bertoni AG, for the MESA Investigators. Association of Asymptomatic Bradycardia With Incident Cardiovascular Disease and MortalityThe Multi-Ethnic Study of Atherosclerosis (MESA). JAMA Intern Med. 2016;176(2):219-227. doi:10.1001/jamainternmed.2015.7655
Bradycardia has been associated with lower cardiovascular disease (CVD) risk in selected populations. There is a paucity of information available about heart rate (HR) less than 50 beats per minute (bpm) among middle-aged or older adults.
To determine whether asymptomatic bradycardia was associated with a lower cardiovascular risk profile, less subclinical atherosclerosis, and decreased incident CVD and mortality.
Design, Setting, and Participants
This retrospective analysis includes 6733 participants of the Multi-Ethnic Study of Atherosclerosis, which recruited men and women free of clinical cardiovascular disease ages 45 to 84 years from 2000 to 2002 and followed them over 10 years for incident CVD events and mortality. The HR was measured by baseline electrocardiogram. The analysis was performed in June 2014.
Main Outcomes and Measures
The association between HR categories with CVD events and all-cause mortality were examined using Cox proportional hazards models adjusted for potential confounders and mediators.
The 6733 participants had a mean (SD) age of 62 (10.2) years; 47% were male. The mean (SD) HR was 63 (9.5) bpm among the 5831 participants not taking an HR-modifying drug; 5.3% had an HR lower than 50 bpm. Preliminary results revealed significant interaction for HR categories according to use of HR-modifying drugs for mortality (P = .002); thus, all further analyses were stratified. An HR of less than 50 bpm was not associated with incident CVD in either subgroup (participants taking or not taking HR-modifying drugs). Among participants not taking HR-modifying drugs, the fully adjusted mortality risk was not different for an HR less than 50 bpm (hazard ratio, 0.71 [95% CI, 0.41-1.09]; P = .12) and increased among those with an HR greater than 80 bpm (hazard ratio, 1.49 [95% CI, 1.08-2.05]; P = .01) (reference HR, 60-69 bpm). Among the 902 participants taking HR-modifying drugs there was an elevated mortality risk associated with an HR less than 50 bpm (hazard ratio, 2.42 [95% CI, 1.39-4.20]; P = .002) and with an HR greater than 80 bpm (hazard ratio, 3.55 [95% CI, 1.65-7.65]; P = .001) (reference HR, 60-69 bpm).
Conclusions and Relevance
In a contemporary, community-based cohort, bradycardia was generally not associated with incident CVD or mortality except for a potential adverse association between bradycardia among those taking HR-modifying drugs.
Bradycardia has traditionally been defined as a heart rate (HR) of less than 60 beats per minute (bpm) or an HR of less than 501- 3 bpm. Multiple studies demonstrate that higher resting HR (RHR) predicts poor cardiovascular outcomes both independent of conventional risk factors and within subgroups of patients with cardiovascular disease.4- 11 Resting heart rate has been demonstrated to be modifiable over time with respect to the interaction of genes12 and environmental factors such as exercise,13,14 medical conditions and medications. However, the impact of bradycardia is less clear. Bradycardia often is found in athletic adults and is typically asymptomatic.15 Among nonathletes, whether persons with bradycardia have higher levels of cardiovascular fitness, or abnormalities in the conduction system is unclear. Particularly among the elderly population, some with bradycardia have symptoms (eg, fatigue, syncope) requiring pacemaker implantation. Tresch and Fleg16 examined a sample (n = 96) and reported that sinus bradycardia (HR <50 bpm) in apparently healthy, nonathletic individuals older than 40 years showed no association with cardiovascular morbidity and mortality. Other studies10,17,18 suggested that asymptomatic bradycardia may be associated with modestly reduced cardiovascular disease (CVD) mortality. Kolloch et al19 examined the relationship of RHR to adverse events, including mortality, in patients with coronary artery disease treated with either verapamil or atenolol, and demonstrated a small increase in risk with an HR less than 50 bpm. Nauman et al20 found that the association of change in RHR with ischemic heart disease mortality was not linear (P = .003 for quadratic trend), suggesting that a decrease in RHR showed no general mortality benefit.
We sought to investigate the association between bradycardia and risk factors, subclinical atherosclerosis, incident CVD, and mortality in a contemporary, well-characterized cohort free of clinical CVD at baseline. We hypothesized that asymptomatic bradycardia was associated with a lower CV risk profile, less subclinical atherosclerosis, and decreased incident CVD and mortality.
The Multi-Ethnic Study of Atherosclerosis (MESA) is a population-based cohort of 6814 men and women from 4 ethnic groups (38% were white; 28%, African American; 22%, Hispanic; and 12%, Chinese) aged 45 to 84 years without clinical CVD prior to study baseline (2000-2002). Details regarding MESA’s design and objectives have been published.21 The cohort was selected from 6 regions in the United States: Forsyth County, North Carolina; North Manhattan and the Bronx, New York; Baltimore City and Baltimore County, Maryland; St Paul, Minnesota; Chicago, Illinois; and Los Angeles County, California. The protocol was approved by the institutional review boards of all participating sites, and written informed consent was obtained from participants. Participants with an ECG tracing at examination 1 were included in these analyses (n = 6733). Participants with atrial flutter, atrial fibrillation, or the presence of a pacemaker at baseline on locally read ECGs were excluded from further participation in MESA.
During the baseline examination (2000-2002), standardized questionnaires and standard calibrated devices were used to obtain demographic data, attained level of education, health habits, medical conditions, current prescription medication use, weight, waist circumference, and height. We used the MESA Typical Week Physical Activity Survey for intentional exercise, defined as the sum of walking for exercise, sports and/or dancing, and conditioning in metabolic equivalent (MET) hours per week.21 We also assessed for any vigorous physical activity (defined as MET level >6). Alcohol use was coded as current/former/never. Participants were asked about current or former cigarette smoking. Body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared. Resting seated blood pressure (BP) was measured 3 times using an automated oscillometric sphygmomanometer (model Pro 100; Critikon); the last 2 BP measurements were averaged for analysis. Hypertension (HTN) was defined based on use of an antihypertensive medication or systolic BP of at least 140 mm Hg or diastolic BP of at least 90 mm Hg. Fasting glucose and lipid levels were measured at a central laboratory. We defined diabetes mellitus as either a glucose level of at least 126 mg/dL or use of glucose-lowering medication. The following medication categories were considered HR modifying: digitalis, β-blockers, nondihydropyridine calcium channel blockers, and all other antiarrhythmic drugs. Dihydropyridine calcium channel blockers were not considered HR-modifying drugs in these analyses. Coronary arterial calcium (CAC) was assessed by computed tomography; each participant was scanned twice, and the average Agatston score was used.22,23
A resting standard 12-lead ECG was obtained from fasting participants using a GE/Marquette MAC 1200 electrocardiograph. The HR was obtained from the ECG rhythm tracing. The total number of beats in 10 seconds (variable for each participant) was used to calculate the beats per minute. The ECGs were centrally read and coded at the Epidemiological Cardiology Research Center (EPICARE), Wake Forest School of Medicine (Winston-Salem, North Carolina). The Minnesota24 coding system was used to classify ECG abnormalities. Participants with only minor ECG abnormalities were classified as having “any minor abnormalities,” and participants with major abnormalities with or without coexisting minor abnormalities were classified as having “any major ECG abnormalities.” Minor ECG abnormalities included minor isolated Q/-QS waves, minor isolated ST/-T abnormalities, high R waves, ST segment elevation; incomplete (left and right) bundle branch block, minor QT prolongation (QTi ≥112%), short PR interval, left axis deviation, right axis deviation, frequent ventricular premature beats, and other minor arrhythmias. Major ECG abnormalities included major ventricular conduction defect; definite myocardial infarction (defined as the presence of major Q wave abnormalities); possible myocardial infarction (defined as the presence of minor Q/-QS wave plus major ST/-T abnormalities); major isolated -ST/-T abnormalities; left ventricular hypertrophy; major atrioventricular conduction abnormalities; and major QT prolongation (QT≥116%), pacemaker, and other major arrhythmias.
The cohort was followed for incident cardiovascular events and mortality at 9- to 12-month intervals; a telephone interviewer contacted each participant to inquire about all interim hospital admissions, cardiovascular outpatient diagnoses and procedures, and deaths. In addition, MESA occasionally identified additional events through cohort clinic visits, participant call-ins, medical record abstractions, or obituaries. To verify self-reported diagnoses, MESA requested copies of all death certificates and medical records for all hospitalizations and selected outpatient cardiovascular diagnoses and procedures. MESA also obtained next-of-kin interviews for out-of-hospital cardiovascular deaths. MESA was successful in getting hospital records on an estimated 99% of suspected hospitalized cardiovascular events. Cardiovascular events were adjudicated centrally by a physician panel. For these analyses, CVD was a composite of myocardial infarction, angina, resuscitated cardiac arrest, stroke (not transient ischemic attack), coronary heart disease (CHD) death, stroke death, other atherosclerotic death, or other CVD death. Cause of death was assigned for potential CVD deaths through committee review. For other deaths, the underlying cause was obtained through state or city vital statistics departments (or occasionally the National Death Index).
Based on prior literature, we defined bradycardia as HR less 50 bpm and grouped HR into 10-bpm increments up to HR greater than 80 bpm. Continuous baseline characteristics were summarized using means (standard deviations) and compared using analysis of variance across HR categories. Preliminary results revealed significant interaction for HR categories according to use of HR-modifying drugs for mortality (P < .002); thus, all further analyses were stratified. χ2 Tests assessed the association between HR category and categorical characteristics. Logistic regression models identified baseline factors associated with an HR of less than 50. We computed rates of incident CVD and mortality by HR category and presented 95% CIs. We used Cox regression models to calculate hazard ratios associated with HR categories, picking as the reference group the category with the lowest event rate. Model 1 adjusted for age, sex, race/ethnicity, education, alcohol status, smoking status, and any vigorous physical activity. Model 2 included model 1 variables and added diabetes mellitus, cholesterol level, high-density lipoprotein cholesterol level, lipid-lowering drug use, BMI, creatinine level, and systolic and diastolic BP. Model 3 added HTN medications. Model 4 added CAC and major ECG abnormalities. We also assessed for interactions between HR and sex, race, and vigorous physical activity. We used STATA statistical software (version 12; StataCorp LP) for statistical analyses.
The HR data were available for 6733 participants (mean [SD] age, 62 [10.2] years; 47% were male; 38% were white; 12%, Chinese; 28%, black; and 22%, Hispanic). The mean (SD) HR was 63 (9.5) bpm among participants not taking an HR-modifying drug; 5.3% had an HR of less than 50 bpm; 30.3%, an HR of 50 to 59 bpm; 39.8%, an HR of 60 to 69 bpm; 18.8%, an HR of 70 to 79 bpm; and 5.9%, an HR of 80 or greater. The corresponding statistics for participants taking HR-modifying drugs are as follows: the mean (SD) HR was 60 (9.7) bpm; 11.2% had an HR of less than 50 bpm; 39.1%, an HR of 50 to 59 bpm; 33.9%, an HR of 60 to 69 bpm; 11.5%, an HR of 70-79 bpm; and 4.2%, an HR of 80 bpm or greater. Baseline characteristics for the sample are depicted in Table 1 and Table 2 according to use of HR-modifying drugs.
Table 3 shows factors associated with an HR of less than 50 stratified by HR-modifying drug use. First, participants taking HR-modifying drugs (Table 2) compared with those not taking HR-modifying drugs (Table 1) tended to be older (mean age, 66.2 [9.3] years vs 61.5 [10.2] years) and had a higher estimated Framingham Risk score (0.12 [0.09] vs 0.10 [0.09]), reflecting a higher burden of CVD risk factors. A greater proportion had CAC greater than 0 (60.7% vs 48.2%) and were more likely to have major ECG abnormalities (22.1% vs 11.3%). Second, among those not prescribed HR-modifying drugs, participants with an HR of less than 50 bpm (Table 3) were more likely to be male and have some characteristics suggesting lower CVD risk (lower BMI, less likely to have diabetes mellitus, more physically active, lower cholesterol level) but had higher mean creatinine level and also were more likely to have ECG abnormalities. In multivariable analyses (Table 3), the following remained independently associated with an HR of less than 50 bpm: male sex, black or Hispanic race/ethnicity, any vigorous physical activity, major ECG abnormalities, diabetes, lower BMI, former alcohol use, graduate/professional education, higher creatinine level and systolic BP, lower diastolic BP, and lower level of triglycerides. Prevalence of CAC was not associated with an HR of less than 50 bpm.
Among the subsample taking HR-modifying drugs, nearly all were prescribed HTN medication. Most were taking β-blockers; only 7% were taking either digitalis or other antiarrhythmic drugs (Table 2). Bradycardia was associated with being male, using β-blockers, and lower mean BMI. Those with an HR of less than 50 bpm were also less likely diabetic. After multivariable adjustment, only male sex and BPs remained statistically significantly associated with an HR of less than 50 bpm (Table 3).
During a median follow-up of 10.1 years (maximum, 11.4 years), 633 incident CVD events occurred (a rate of 10.3 per 1000 person-years). The incidence of CVD was lowest in the HR category of 50 to 59 bpm in both subgroups, and highest in those with an HR greater than 80 bpm. The incidence of CVD was generally higher in participants taking HR-modifying drugs in all HR categories (see the eFigure in the Supplement). However, a test for an interaction between HR category and HR-modifying drug use with CVD events was not significant (P = .33). Table 4 depicts adjusted hazard ratios for CVD using those with an HR of 50 to 59 bpm as the reference group. After adjustment, the slight excess of CVD events among those with an HR of less than 50 bpm was not significant (P = .74). In contrast, CVD risk was consistently statistically significantly elevated for those with HR 80 and above.
During follow-up, there were 697 deaths (a rate of 10.5 per 1000 person-years); 160 of these were CVD-related deaths, and 537 were non–CVD-related deaths. The mortality rate is depicted in the Figure. Mortality was significantly higher in individuals with an HR of less than 50 bpm (P = .001) and an HR greater than 80 bpm (P < .001) in patients who used HR-modifying drugs. A U-shaped distribution for mortality rate is evident in patients prescribed HR-modifying drugs. In contrast, for participants not taking HR-modifying drugs the pattern appeared more linear. For all mortality analyses a P test for trend was significant.
Table 4 depicts hazard ratio calculations for mortality using the 4 models defined in the Statistical Analysis section. The reference was an HR of 60 to 69 bpm because mortality was the lowest in this group. Notably, there seemed to be a linear increase in hazard ratio with increasing HR categories in patients not prescribed HR-modifying drugs. In individuals taking HR-modifying drugs, there was a statistically significant increase in hazard ratio in participants with an HR of less than 60 and an HR greater than 80 in a J-shaped distribution. Most notably, the fully adjusted hazard ratios for mortality in individuals taking HR-modifying drugs for RHR of less than 50 bpm and HR of 50 to 59 bpm were 2.42 (95% CI, 1.39-4.20; P = .002) and 1.65 (95% CI, 1.05-2.59; P = .03), respectively.
We did not find significant interactions by sex, race/ethnicity, or vigorous physical activity. We conducted 2 sensitivity analyses showing consistent patterns. The first sensitivity analysis defined a healthy cohort not taking HR-modifying drugs, without HTN, and without major ECG abnormalities. The second sensitivity analysis defined a group of participants taking HR-modifying drugs but not taking digoxin or other antiarrhythmic drugs.
In this cohort of middle-aged and older adults who were free of clinically apparent CVD at baseline, resting bradycardia was not uncommon. The crude incidence of CVD was higher in those with bradycardia, but with adjustment for CVD risk factors and potential confounders, this excess risk was not significant. The crude mortality rate was higher in those with bradycardia; however, this was explained by differences in mortality according to the use of HR-modifying drugs. Consistent with findings in some prior literature, we did not find an excess mortality associated with bradycardia in the sample of individuals not taking HR-modifying drugs. Mortality in individuals taking HR-modifying drugs was 69% higher in those with an HR of 50 to 59 bpm and 142% higher in those with an HR of less than 50 bpm compared with an HR of 60 to 69 bpm. Finally, our results are consistent with those of prior literature that demonstrates that the incidence of CVD and mortality are higher with an increasing HR greater than 80 bpm and with a prior MESA finding of increased heart failure with higher HR.25
The finding of higher mortality among participants free of clinical CVD with an HR of less than 50 bpm for participants taking HR-modifying drugs was somewhat unexpected. The mortality in individuals taking HR-modifying drugs was significantly higher in those with HRs of less than 60 bpm and greater than 80 bpm, consistent with a J-shaped curve. For those not prescribed HR-modifying drugs, the curve appears more linear. Although there was increased mortality in people not taking HR-modifying drugs in those with higher HRs, these results seem to be attenuated secondary to adjustment for both CAC and major ECG abnormalities. In addition, sensitivity analyses revealed the same pattern when excluding participants with major ECG abnormalities.
The literature relating high RHRs (tachycardia) to CVD and mortality is compelling. Our results are generally concordant with those in the current literature in participants not taking HR-modifying drugs. In contrast, Paul et al17 demonstrated a linear relationship between RHR and mortality in normotensive patients and those with untreated HTN. Jouven et al9,10 studied approximately 5700 working middle-aged men without known or suspected cardiovascular disease; lower baseline HR was not associated with an increased mortality risk. Diaz et al5 assessed the relationship between RHR in patients with suspected or proven CAD with median follow-up for 14.7 years and found that all-cause mortality, cardiovascular mortality, and cardiovascular rehospitalizations were increased with increasing HR. In our study, we assessed patients without any known CVD but found similar increased CVD and mortality with higher HR.
There are relatively few studies reporting the association of low RHR and mortality. Most of the literature discusses HR change over time.11,17,19,20 Two studies found a relatively linear relationship between heart rate and mortality. Okin et al11 found a linear relationship between in-treatment HR on serial ECGs with CVD and mortality. The study11 specifically examined hypertensive patients with LVH and included individuals with DM, CAD, stroke, PVD, atrial fibrillation, and heart failure. Paul et al17 examined change in HR over time in a hypertensive cohort without pacemakers or atrial fibrillation. Survival analysis adjusted for age, sex, BMI, smoking, rate-limiting therapy, systolic BP, and serum cholesterol level and concluded that there was little indication of a J-shaped relationship for final HR or change in HR.
The literature has previously demonstrated J-shaped relationships between HR and outcomes. Kolloch et al19 specifically examined BP control and adverse outcomes using either a verapamil sustained release–based or an atenolol-based treatment strategy were evaluated in 22 192 patients with HTN 50 years or older with clinically stable CAD (without pacemakers). First, in this study baseline (pretreatment) bradycardia was not associated with adverse outcomes after adjustments; however, follow-up HR showed a J-shaped relationship more marked in those with previous myocardial infarction or diabetes mellitus, suggesting that excessive HR reduction may be harmful. Nauman et al20 conducted a prospective study that examined HR change over time. The analyses excluded individuals with current use of BP drugs, history of angina pectoris, myocardial infarction, diabetes mellitus, or stroke. Similar to our results, in their study the association of change in RHR with ischemic heart disease mortality was not linear, suggesting that a decrease in RHR showed no general mortality benefit. Finally, the recent SIGNIFY trial26 randomizing adults with CAD to ivabradine or placebo failed to show that lowering the HR (mean HR, 60.7 vs 70.6 bpm) reduced the incidence of CVD overall, and found a subgroup of participants (those with activity-limiting angina) for whom ivabradine use was associated with an increased risk of cardiovascular death and nonfatal myocardial infarction.26
There are some limitations to these findings, including the potential for confounding by indication because we could not determine why individuals were taking HR-modifying drugs at baseline. It is possible that these individuals may have been using these drugs for arrhythmias, heart failure, or other CVD not reported to MESA. Perhaps more likely, persons taking these drugs may have had a previously higher BP and that adjusting for a single, baseline BP may not fully account for the increased risk associated with high BP. Thus, the adverse hazard noted for individuals taking these drugs reflects not the drug effect or HR, but rather the previously more severe HTN relative to other participants. The design of MESA precludes determining participants’ HR prior to treatment with drugs that might affect HR. In addition, our subgroup analyses created smaller sample sizes, and the small number of CVD-related deaths precluded a separate analysis. However, there are several strengths to these analyses as well. MESA is a well-characterized, sex-balanced and ethnically diverse population-based sample with multiyear follow-up for events, which were adjudicated. In contrast with several prior studies, MESA was a CVD-free population. We were also able to adjust for multiple potential confounders and comorbidities as well as for subclinical atherosclerosis.
Our results may be reassuring to most adults found to have asymptomatic bradycardia. In contrast, the association of bradycardia with mortality among participants prescribed drugs that may slow HR may have clinical relevance. Further work is needed in other cohorts and trial databases to see whether these findings can be replicated and if confirmed, to determine whether this association is causally linked to HR or to use of these agents.
Corresponding Author: Ajay Dharod, MD, Department of General Internal Medicine, Wake Forest School of Medicine, 1 Medical Center Blvd, Winston Salem, NC 27157 (firstname.lastname@example.org).
Accepted for Publication: November 17, 2015.
Published Online: January 19, 2016. doi:10.1001/jamainternmed.2015.7655.
Author Contributions: Drs Dharod and Bertoni had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Soliman, Dawood, Bertoni.
Acquisition, analysis, or interpretation of data: Dharod, Soliman, Chen, Shea, Nazarian, Bertoni.
Drafting of the manuscript: Dharod, Bertoni.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Dharod, Bertoni.
Obtained funding: Shea.
Administrative, technical, or material support: Nazarian.
Study supervision: Soliman, Shea, Bertoni.
Conflict of Interest Disclosures: Dr Nazarian is principal investigator for research funding to Johns Hopkins University from Biosense Webster Inc. He is also a scientific advisor to Biosense Webster, Medtronic, and CardioSolv, Inc. No other disclosures are reported.
Funding/Support: This research was supported by contracts N01-HC-95159, N01-HC-95160, N01-HC-95161, N01-HC-95162, N01-HC-95163, N01-HC-95164, N01-HC-95165, N01-HC-95166, N01-HC-95167, N01-HC-95168 and N01-HC-95169 from the National Heart, Lung, and Blood Institute and by grants UL1-TR-000040 and UL1-TR-001079 from NCRR. Dr Shea has received grant support from NIH as noted in these contracts.
Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: The authors thank the other investigators, the staff, and the participants of the MESA study for their valuable contributions. We also thank Charles N. Campbell Jr, BS AAS, at Wake Forest School of Medicine for assistance with ECG data. Mr Campbell received no financial compensation for his contribution.