Article Text

Download PDFPDF

Original research article
Resting heart rate, temporal changes in resting heart rate, and overall and cause-specific mortality
  1. Mathias Seviiri1,
  2. Brigid M Lynch1,2,3,
  3. Allison M Hodge1,2,
  4. Yi Yang1,2,
  5. Danny Liew4,
  6. Dallas R English1,2,
  7. Graham G Giles1,2,
  8. Roger L Milne1,2,
  9. Pierre-Antoine Dugué1,2
  1. 1 Cancer Epidemiology and Intelligence Division, Cancer Council Victoria, Melbourne, Victoria, Australia
  2. 2 Centre for Epidemiology and Biostatistics, Melbourne School of Population and Global Health, The University of Melbourne, Melbourne, Victoria, Australia
  3. 3 Physical Activity Laboratory, Baker Heart and Diabetes Institute, Melbourne, Victoria, Australia
  4. 4 School of Public Health and Preventive Medicine, Monash University, Melbourne, Victoria, Australia
  1. Correspondence to Dr Pierre-Antoine Dugué, Cancer Epidemiology and Intelligence Division, Cancer Council Victoria, Centre for Epidemiology and Biostatistics, Melbourne School of Population and Global Health, The University of Melbourne, Melbourne, Victoria 3004, Australia; pdugue{at}cancervic.org.au

Abstract

Objective Most studies investigating the association between resting heart rate (RHR) and mortality have focused on cardiovascular disease (CVD) mortality, and measured RHR at only one time point. We aimed to assess associations of RHR and changes in RHR over approximately a decade with overall and cause-specific mortality.

Methods We used data from participants in the Melbourne Collaborative Cohort Study with RHR measures at baseline (1990–1994; n=41 386; 9846 deaths) and at follow-up (2003–2007; n=21 692; 2818 deaths). RHR measures were taken by trained staff, using Dinamap monitors. Cox models were used to estimate HR and 95% CI for the associations between RHR and mortality. Vital status and cause of death were ascertained until August 2015 and December 2013, respectively.

Results After adjustment for confounders, including blood pressure and known medical conditions but not arrhythmias or atrial fibrillation, RHR was associated with a higher risk of death of similar magnitude for CVD (HR per 10 beats per minute (bpm)=1.11, 95% CI 1.07 to 1.16), cancer (HR=1.10, 95% CI 1.06 to 1.13) and other causes (HR=1.20, 95% CI 1.16 to 1.25). Higher mortality was observed for most cancer sites, including breast (HR=1.16, 95% CI 1.03 to 1.31), colorectal (HR=1.18, 95% CI 1.08 to 1.29), kidney (HR=1.27, 95% CI 1.03 to 1.57) and lung cancer (HR=1.19, 95% CI 1.10 to 1.29). Temporal increases in RHR were associated with higher mortality, particularly for individuals whose RHR increased by more than 15 bpm.

Conclusions RHR and changes in RHR over a decade are associated with mortality risk, including from causes other than CVD such as breast, colorectal or lung cancer. Monitoring of RHR may have utility in identifying individuals at higher mortality risk.

  • cardiac risk factors and prevention
  • epidemiology

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Resting heart rate (RHR) is a non-invasively measured marker of cardiac function and a good indicator of physical fitness and general health.1 Elevated RHR measured at a single point in time has been consistently shown to be associated with increased all-cause and cardiovascular disease (CVD) mortality.2–10 The evidence on the association between RHR and cancer mortality is less conclusive. No association was found between RHR and cancer mortality among middle-aged Finnish adults10 or among Israeli men.4 Other studies have found that elevated RHR is associated with all-cancer mortality both in the general population11 and among men with CVD.10 12 Mortality from specific cancers has rarely been investigated, probably because RHR has seldom been measured in large cohort studies. In studies of modest size, higher RHR was found to be associated with poorer breast cancer survival13 and higher prostate cancer mortality.14

Measuring RHR at two time points may provide insight into unmeasured changes relating to lifestyle and physical fitness.15 16 Available data suggest that an increase in RHR over time is associated with higher all-cause mortality in people with and without CVD.6 17–19 The evidence on whether a decrease in RHR over time is associated with lower mortality is inconsistent.17–19 Among the few studies that have assessed the association between temporal changes in RHR and all-cause mortality, some failed to adjust for important confounders like physical activity, alcohol consumption and diet; a meta-analysis found that the association of RHR with cancer mortality was almost twofold higher in less adjusted studies, compared with those including more confounders.12 Some studies were sex-specific or disease-specific,7 19 and others assessed only short-term RHR changes (eg, over weeks/months).19

The aim of this study was to examine the associations of RHR measured at a single time point and of RHR changes over time with all-cause and cause-specific mortality among participants in a large prospective cohort study.

Methods

The Melbourne Collaborative Cohort Study (MCCS) is a prospective cohort study that included 41 514 adult women (24 469; 59%) and men (17 045; 41%) aged 40–69 years (for 99% of them) at baseline. Participants were from the Melbourne metropolitan area, and were recruited between 1990 and 1994, at which time information on demographics, lifestyle factors and medical conditions was collected through questionnaires, and clinical measurements were taken by trained staff. Greek-born and Italian-born participants were oversampled to extend the range of dietary, lifestyle and genetic characteristics of the population. A follow-up survey conducted between 2003 and 2007 (wave 2) repeated most baseline measures.20 Written informed consent was provided by all participants.

Exposure measurement

At both baseline and wave 2, RHR (measured as beats per minute; bpm) was assessed using a Dinamap 1846SX automatic blood pressure monitor (Critikon, Tampa, Florida, USA). Three consecutive readings were taken in a sitting position at 1 min intervals, after 5 min rest, by trained personnel. The average of the second and third readings was computed to obtain an accurate RHR measure. The presence of arrhythmias and atrial fibrillation could not be assessed. Diastolic and systolic blood pressures were measured at both time points using a similar protocol.

Measurement of other variables

Demographic variables were obtained by questionnaires at baseline and wave 2, and included age, sex, country of birth and highest level of education attained (table 1). Lifestyle factors included smoking status and intensity, alcohol consumption, and physical activity (summary score based on the relative energy expenditure21). A 121-item Food Frequency Questionnaire (FFQ) was developed specifically for the MCCS.22 The Alternative Healthy Eating Index (AHEI-2010) was used as a measure of diet quality.23–25 Other dietary factors considered included caffeine intake, and the ratio of sodium intake to potassium intake, both derived from the FFQ. During baseline and wave 2 interviews, history of the following medical conditions was recorded: stroke, heart attack, angina, hypertension, diabetes and asthma. Clinical measurements included total serum cholesterol and waist circumference.

Table 1

Characteristics of participants in the Melbourne Collaborative Cohort Study (n=41 386), by RHR

Outcome ascertainment

Mortality was ascertained through record linkage to the Victorian Registry of Births, Deaths and Marriages, the National Death Index, and Victoria Cancer Registry. Cause-specific mortality was defined according to the WHO International Classification of Diseases codes versions ICD-9 and ICD-10, using the underlying cause of death from the death certificate. The cause was unknown for 157 deaths and classified as ‘other cause’.

Exclusion criteria

Out of 41 514 baseline participants, 128 were excluded due to missing baseline RHR values. For the RHR change analysis, the following participants were excluded: died (n=5974), left Australia (n=86), or did not attend wave 2 or their RHR was not recorded (n=13 720). A total of 21 692 participants were included in the analysis of temporal change in RHR (online supplementary figure 1).

Supplementary file 1

Statistical analysis

Cox regression models were used to estimate HRs and 95% CIs for the associations between RHR and mortality. Attained age was used as the time metric.26 For baseline analyses, follow-up began at the date of recruitment, while for RHR change analyses it began on the date of wave 2 attendance. Follow-up ended at date of death, date participants left Australia or date of the latest data linkage (31 August 2015 for all-cause mortality, and 31 December 2013 for cause-specific mortality, dates at which data were considered complete), whichever came first. The proportional hazards assumption was evaluated using Schoenfeld residuals and, for all baseline and change in RHR measures, there was no evidence of violation.

In the first analysis, the association between baseline RHR and all-cause and cause-specific mortality (all-cancers, all CVD and all other non-CVD/non-malignancy) was assessed. Baseline RHR was categorised as <60 bpm, 60–69 bpm, 70–79 bpm, 80–89 bpm and 90+ bpm, consistent with previous publications.3 4 15 The likelihood ratio test was used to test for a linear association between baseline RHR and all-cause mortality by comparing models with baseline RHR as a categorical and pseudo-continuous variable. We conducted stratified analyses by sex, and tested the interaction between sex and RHR using likelihood ratio tests. In the second analysis, the associations between change in RHR (difference between wave 2 and baseline RHR) and all-cause and cause-specific mortality were assessed. The temporal change in RHR was categorised as very large decrease (more than 25 bpm), large decrease (16–25 bpm), moderate decrease (6–15 bpm), stable RHR (−5 to +5 bpm), moderate increase (6–15 bpm), large increase (16–25 bpm) and very large increase (more than 25 bpm). The above categories were applied a priori as being potentially clinically relevant, consistent with a previous publication.16 Participants with a stable RHR were used as the reference group. In the third analysis, change in RHR was grouped by clinical status at baseline and wave 2; an RHR of 70 bpm or less was considered normal.7 14 15 Participants were grouped into four categories: those maintaining a healthy RHR between baseline and wave 2 (normal to normal), those with deteriorating RHR (normal to high), those improving RHR (high to normal) and those maintaining an unhealthy RHR (high to high).

HRs are reported for minimally adjusted (age and sex) and multivariable-adjusted models using a directed acyclic graph to represent the confounders we considered27 (online supplementary figure 2). All potential confounders were included in the models, all measured at baseline: sex, country of birth, educational attainment, waist circumference, alcohol consumption, smoking, physical activity score, blood pressure, total serum cholesterol, AHEI-2010, caffeine intake, sodium:potassium intake ratio, and history of asthma, angina, hypertension and diabetes. In addition to the above potential confounders, wave 2 waist circumference and baseline RHR were included in the RHR change analysis (online supplementary figure 2).

We tested whether the association between change in RHR and all-cause mortality differed by sex, age at wave 2 (dichotomised at 65 years), physical activity (low vs high) and hypertension status by comparing models with and without interactions using likelihood ratio tests. As the association between RHR and mortality may be affected by pre-existing clinical conditions (reverse causation), we conducted sensitivity analyses excluding the first 2 years of follow-up and excluding individuals with reported cardiovascular conditions at baseline. All statistical analyses were performed using Stata V.14.1.

Results

Baseline and wave 2 characteristics of the participants are summarised in table 1. During a median follow-up of 21.9 years, 9846 deaths (23.8%) were observed. For the analysis of cause-specific mortality (median follow-up of 20.3 years), there were 3618 cancer deaths, 2239 CVD deaths and 2573 deaths from other causes. Elevated baseline RHR was associated with higher risk of death from all causes, CVD, cancer and other causes, after adjusting for potential confounders (table 2). These associations were consistent with a linear trend (all-cause: HR per 10 bpm increase=1.13, 95% CI 1.11 to 1.15; CVD: HR=1.11, 95% CI 1.07 to 1.16; cancer: HR=1.10, 95% CI 1.06 to 1.13; and other-cause mortality: HR=1.20, 95% CI 1.16 to 1.25). Results were consistent after excluding the first 2 years of follow-up for both minimally adjusted models (not shown) and fully adjusted models (table 2). Similar results were obtained after excluding participants with reported pre-existing cardiovascular conditions (online supplementary table 5). We found weak evidence of interaction between sex and RHR (P interaction=0.07 for overall mortality; figure 1, online supplementary table 1). Higher site-specific cancer mortality was observed, particularly for breast (HR per 10 bpm=1.16, 95% CI 1.03 to 1.31), colorectal (HR=1.18, 95% CI 1.08 to 1.29), kidney (HR=1.27, 95% CI 1.03 to 1.57) and lung cancer (HR=1.19, 95% CI 1.10 to 1.29) (table 3). Similar associations were observed in men and women (figure 2, online supplementary table 2).

Supplementary file 2

Figure 1

Association of resting heart rate measured at baseline with all-cause and cause-specific mortality, stratified by sex. Cox models were adjusted for age, country of birth, level of education, waist circumference, alcohol consumption, smoking, physical activity score, alternate healthy eating index, total serum cholesterol, sodium-potassium ratio, caffeine, blood pressure, history of hypertension, angina, asthma and diabetes. Interactions between sex and resting heart rate (RHR) were tested by comparing models with and without interaction terms between sex and RHR, using likelihood ratio tests. P-values for interaction: all-cause mortality: P=0.07, in the 90+bpm category: P=0.02; cancer mortality: P=0.72; CVD mortality: P=0.77; other-cause mortality: P=0.55. bpm, beats per minute; CVD, cardiovascular disease.

Figure 2

Association of resting heart rate (per 10 bpm) measured at baseline with mortality by cause, stratified by sex. Cox models were adjusted for age, country of birth, level of education, waist circumference, alcohol consumption, smoking, physical activity score, alternate healthy eating index, total serum cholesterol, sodium-potassium ratio, caffeine, blood pressure, history of hypertension, angina, asthma and diabetes. Hazard ratios are expressed for a 10bpm increase in the resting heart rate variable. Results for male breast cancer are not shown because based on a too small number of deaths (N=3, HR= 0.85, 95% CI: 0.23-3.07, P=0.80). Bpm, beats per minute; CVD, cardiovascular disease; UADT, upper aerodigestive tract.

Table 2

Baseline RHR and mortality in the Melbourne Collaborative Cohort Study (n=41 386)

Table 3

Baseline RHR and cause-specific mortality risk in the Melbourne Collaborative Cohort Study (n=41 386)

The change in RHR analysis included 21 692 participants (12 967 (60%) women and 8725 (40%) men) with a mean RHR of 65 bpm (SD=10 bpm) at wave 2. There was a mean decrease in RHR of 3 bpm (SD=10) over time. During a median follow-up of 10.8 years following wave 2, there were 2818 deaths (13%) from all causes. Compared with a stable RHR, an increase in RHR was associated with higher overall mortality risk, but not decreases in RHR (table 4). Similar results were found for CVD and other-cause mortality (table 5). The association between changes in RHR and all-cause mortality varied by sex (P=0.02; figure 3, online supplementary table 3) but not by history of hypertension (P interaction=0.9), physical activity (P=0.3) or age at wave 2 (P=0.3). Online supplementary table 4 shows the associations between change in RHR and cause-specific death.

Figure 3

Association of change in RHR from baseline to wave 2 with all-cause mortality, stratified by sex. Cox models were adjusted for age, country of birth, level of education, waist circumference (baseline & wave2), baseline RHR, alcohol consumption, smoking, physical activity score, AHEI, total serum cholesterol, sodium-potassium ratio, caffeine, blood pressure, hypertension, angina, asthma and diabetes. Interaction between sex and resting heart rate (RHR) was tested by comparing models with and without interaction terms between sex and RHR, using likelihood ratio tests. P-interaction = 0.02. Bpm, beats per minute; RHR, resting heart rate.

Table 4

Temporal changes in resting heart rate (RHR) and all-cause mortality in the Melbourne Collaborative Cohort Study (n=21 692)

Table 5

Temporal changes in RHR and cause-specific mortality in the Melbourne Collaborative Cohort Study (n=21 692)

When considering broader, clinically focused categories of change in RHR, compared with participants who maintained a healthy RHR between baseline and wave 2, individuals whose RHR increased and those with a consistently unhealthy RHR had higher all-cause mortality, after adjusting for potential confounders. Participants whose RHR decreased from high to normal had similar risk to those who maintained a healthy RHR (table 4).

Results from sensitivity analyses (excluding participants with pre-existing clinical conditions) were not materially different (online supplementary table 5).

Discussion

In this large, prospective cohort study of adult men and women, the association of RHR with cancer and other-cause mortality was of similar magnitude to that of CVD mortality. In addition, an increase in RHR over approximately a decade was associated with a markedly higher all-cause mortality risk. While a decrease in RHR did not appear to be associated with any of the outcomes considered, deteriorating and consistently high RHRs over time were associated with higher risks of death, compared with participants maintaining a normal RHR.

The strengths of our study include its prospective design, the large number of events (9846 for the baseline and 2818 deaths for the change analysis, respectively), and a long follow-up period for both baseline and wave 2 analyses (up to 25 and 12 years, respectively). RHR measures were taken by trained staff using validated monitors. The change in RHR was measured over a decade, thereby potentially measuring true changes related to physical fitness rather than short-term intraindividual variability. We adjusted for a large range of potential confounders including physical activity, alcohol consumption, dietary factors, chronic conditions and metabolic factors such as blood pressure, which was not systematically done in previous studies. Nevertheless, given the observational nature of our study, we cannot completely rule out that some of the observed associations were due to unmeasured confounding. We conducted sensitivity analyses that excluded the first 2 years of follow-up, or participants with CVD conditions, showing that the observed associations were unlikely due to reverse causation. These analyses further strengthened our conclusion of an independent role played by RHR in the risk of death.

Study limitations include a relatively large attrition between study time points (24% of participants did not attend wave 2 clinic so might have been less healthy than the baseline sample, and 15% did not have RHR recorded because no monitor was available at the clinical assessment). Another important limitation is the lack of information on medication at wave 2 that may alter RHR (such as beta-blockers), which is a strong potential confounder. Hence, we could not examine how these affected the associations, as has been done in other studies.17 Although these studies did not point to a strong confounding effect of such medication, this may to a certain extent explain the pattern we observed for participants with decreasing RHR over time: these individuals may have been medically treated due to poor health rather than becoming physically fitter during follow-up. Our observation of a slight decrease in RHR over a decade, although in line with other studies,18 28 requires further investigation. Lastly, we had no data on arrhythmias or atrial fibrillation, which may cause elevated RHR and increase the risk of death. While these are important missing variables, such confounding is unlikely to have explained all the observed association between RHR and mortality.

Some aspects of our findings are consistent with previous studies. In the Nord-Trøndelag Health Study study of 13 499 men and 15 826 women in Norway,18 decreases in RHR over a 10-year period were not associated with all-cause and ischaemic heart disease mortality. The authors reported a 20% higher all-cause mortality risk for participants whose RHR increased from less than 70 bpm to 70–85 bpm. In our study, the risk was higher by 24% in participants whose RHR deteriorated from 70 bpm or less to over 70 bpm, compared with those who maintained a healthy RHR of 70 bmp or less between baseline and wave 2. In the Nord-Trøndelag Health Study study, participants who maintained a high RHR of 70 bpm or higher had a 30% increased all-cause mortality, compared with those who maintained an RHR of less than 70 bpm. We observed a 16% higher risk for comparable elevated RHR. In terms of CVD mortality, the Nord-Trøndelag Health Study (HUNT)  found an increase in RHR greater than 25 bpm, compared with a stable RHR (−5 to +5 bpm), was associated with an 80% higher ischemic heart disease mortality.18 Consistent with this, we observed large mortality increases for individuals with large RHR increases. In the Paris Prospective Study 1, which included 15 139 healthy working men aged 42–53 years, the second, third and fourth quartiles (compared with the first quartile) of RHR were associated, respectively, with 10%, 20% and 70% higher all-cause mortality over 5 years.11 Earlier findings from the same study showed that an increase of more than 3 bpm in RHR over 5 years was associated with a 19% higher all-cause mortality compared with stable RHR, and that a decrease of more than 4 bpm was associated with a 14% decreased all-cause mortality.17 A decrease in RHR in our study was not associated with lower mortality, contrary to the reported findings from the Paris Prospective Study 1. However, the Paris Prospective Study 1 results were not adjusted for potentially important confounders such as alcohol consumption and dietary factors, and were also restricted to men.17 18 Thus, their results may not be directly comparable to ours. The Candesartan in Heart Failure-Assessment of Reduction in Mortality and Morbidity programme in the USA19 included 7599 participants with chronic heart failure and reported a 9% higher all-cause mortality for every 5 bpm increase in RHR between clinic visits, and a 15% lower risk for a 10 bpm decrease in RHR compared with no change in RHR. This study only considered patients with chronic heart failure and a short-term RHR change (median follow-up of 3 years). Our study therefore adds to the scarce literature on temporal changes in RHR and mortality, and supports the hypothesis that increased RHR over time is associated with substantially higher mortality risk. In line with previous studies, we did not find that decreased RHR over time was associated with lower mortality,17 19 which suggests the importance of maintaining a healthy cardiometabolic profile.

Our findings concur with the results of studies showing elevated RHR measured at a single point to be associated with higher CVD mortality.2 12 Our study also adds to the evidence that elevated RHR measured at a single point is associated with cancer mortality, for which evidence has been less consistent.4 10 12 Our analyses also suggest a higher mortality for certain cancer sites, such as breast, colorectal and lung for which, to our knowledge, very little evidence exists in the literature,13 14 probably owing to the fact that few large cohort studies have routinely collected RHR.

High levels of sympathetic activity, characterised by an elevated RHR, increase adrenergic activity stimulation, which leads to increased neurotrophic factors that stimulate epithelial cell growth and proliferation, potentially resulting in disease such as cancer.29 The initiation and progression steps involve processes such as inflammation, impaired cellular response and transition of the epithelial mesenchyme.30 Additionally, high RHR triggers deposition of atherosclerotic plaques by exerting mechanical stress on the walls of the arteries, which contribute to CVD mortality.30 31 Elevated RHR may also reflect underlying arrhythmias, which increase mortality risk.32 Although RHR has been thought to be a silent bystander,33 a recent genome-wide association study has confirmed the causal role of genetically determined RHR in mortality,34 which found relative risks larger than those of our study. Few data exist exploring the association between RHR and the risk of cancer. Given the higher mortality we observed for cancer, it seems warranted to disentangle whether these were driven by higher cancer risk or poorer survival.

Conclusion

Elevated RHR measured at a single point is associated with higher risk of death from most causes, including cancer, independently of blood pressure and the main sociodemographic and lifestyle risk factors. Long-term increases in RHR were also associated with higher mortality. RHR monitoring during general practitioner visits or using fitness devices could help identify population subgroups at higher mortality risk.

Key messages

What is already known on this subject?

Resting heart rate (RHR) is an objective and non-invasively measured marker of physical fitness and general health. Most existing studies focused on cardiovascular mortality and measured RHR at only one time point.

What might this study add?

After adjustment for potential confounders, RHR was associated with substantially higher mortality, including from causes such as breast, colorectal and lung cancer. Increases in RHR measured over a decade were also associated with higher mortality, particularly for individuals whose RHR increased by more than 15 bpm.

How might this impact on clinical practice?

Monitoring of RHR through health checks or using fitness devices may have utility in identifying individuals at higher mortality risk.

Acknowledgments

Cases and their vital status were ascertained through the Victorian Cancer Registry (VCR) and the Australian Institute of Health and Welfare (AIHW), including the National Death Index and the Australian Cancer Database. This study was made possible by the contribution of many people, including the original investigators and the diligent team who recruited the participants and who continue working on follow-up. We express our gratitude to the thousands of Melbourne residents who took part in the study.

References

Footnotes

  • Contributors Study concept and design: MS, BML, YY, P-AD. Acquisition of data: DRE, GGG. Analysis and interpretation of data: MS, BML, AMH, YY, DL, DRE, GGG, RLM, P-AD. Drafting of the manuscript: MS, BML, P-AD. Critical revision of the manuscript for important intellectual content: MS, BML, AMH, YY, DL, DRE, GGG, RLM, P-AD. Statistical analysis: MS, P-AD. Study supervision/guarantor: P-AD. All authors approved the final manuscript submitted and they approved the authorship list.

  • Funding The MCCS cohort recruitment was funded by VicHealth and Cancer Council Victoria. The MCCS was further supported by the Australian NHMRC grants 209057 and 396414, and by infrastructure provided by Cancer Council Victoria. BML is supported by an NBCF Fellowship (ECF 15-012).

  • Competing interests None declared.

  • Ethics approval The Human Research Ethics Committee at Cancer Council Victoria approved the study protocol.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement Enquiries about statistical code can be directed to the corresponding author.

Linked Articles