Resting Heart Rate Pattern During Follow-Up and Mortality in Hypertensive Patients
There is a linear relationship between resting heart rate (HR) and mortality in normotensive and untreated hypertensive individuals. However, it is not clear whether HR is a marker of increased risk in hypertensive patients on treatment. We investigated the relationship between HR and mortality in patients with hypertension. We analyzed baseline HR, final HR, and HR change during follow-up in patients attending the Glasgow Blood Pressure Clinic. Using a threshold of 80 bpm, we classified patients into those who had a consistently high (high-high) or low (low-low) HR or patients whose HR increased (low-high) or decreased (high-low) over time. Survival analysis was carried out using Cox proportional hazards models adjusted for age, sex, body mass index, smoking, rate-limiting therapy, systolic blood pressure, and serum cholesterol. For each beat of HR change there was a 1% change in mortality risk. The highest risk of an all-cause event was associated with patients who had increased their HR by ≥5 bpm at the end of follow-up (1.51 [95% CI: 1.03 to 2.20]; P=0.035). Compared with low-low patients, high-high patients had a 78% increase in the risk of all-cause mortality (HR: 1.78 [95% CI: 1.31 to 2.41]; P<0.001). Cardiovascular mortality showed a similar pattern of results. Rate-limiting therapy did not have an independent effect on outcomes in this analysis. Change in HR achieved during follow-up of hypertensive patients is a better predictor of risk than baseline or final HR. After correction for rate-limiting therapy, HR remained a significant independent risk factor.
There is evidence to suggest a prognostic role for resting heart rate (HR) in patients with coronary artery disease, heart failure, myocardial infarction, and hypertension.1–7 In patients with myocardial infarction and heart failure, reduction in HR by β-blockers is associated with a reduction in cardiovascular (CV) outcomes.7–9 More recently, the BEAUTIFUL (morBidity-mortality EvAlUaTion of the If inhibitor ivabradine in patients with coronary disease and left ventricULar dysfunction) Study showed that reduction in HR with ivabradine did not improve cardiac outcomes in patients with stable coronary artery disease and left ventricular systolic dysfunction, but it did reduce adverse coronary outcomes in a subgroup of patients with HR >70 bpm.10
In hypertension, the association between HR and risk is uncertain. Epidemiological studies have shown an association between high HR and mortality.5 The International Verapamil-SR/Trandolapril Study (INVEST) found follow-up HR to show a J-shaped relationship with adverse outcome in patients with coronary artery disease and hypertension.11 An analysis of the placebo-treated arm of the Systolic Hypertension in Europe Trial identified an association between HR and mortality in elderly patients, although follow-up was relatively short and the predictive power of HR lost significance after adjustment in the treated arm.12 A large meta-analysis13 suggested that reduction in HR (because of β-blockers) did not confer any additional benefit in hypertensive patients. However, it is possible that pharmacological HR lowering with β-blockers has adverse central hemodynamic effects leading to adverse outcomes.14 To further clarify the relation between HR and long-term outcomes in hypertension we studied baseline HR, achieved HR, change in HR, and HR pattern during follow-up in a large hypertensive population.
Subjects used in this analysis were identified from the Glasgow Blood Pressure Clinic. The Glasgow Blood Pressure Clinic provides secondary and tertiary level service to individuals with hypertension from the west of Scotland. Data from patients attending the clinic are stored in a computerized database, which contains information for >11 000 individuals. Record linkage with the office of the Register General for Scotland allows identification of all deaths and causes of death in clinic attendees.
The Glasgow Blood Pressure Clinic uses specialist hypertension nurses who measure HRs manually in beats per minute. Patients are asked to rest for 5 minutes in the supine position before HR is measured over 1 minute, and the average of 3 measurements is recorded. HR is measured between 9:00 am and 11:00 am for all patients before blood pressure measurement and the application of the blood pressure cuff. Blood pressure is manually measured 3 times, using standardized sphygmomanometers; the mean of the last 2 measurements is recorded. We did not have control over any food or drink consumed by the patient before their appointment at the clinic. Patients are advised to take their regular medications as usual. Patients with electronic pacemakers or atrial fibrillation were not included in the analysis. If frequent ectopic beats >5 per minute were noted during measurement, then that reading was not included for analysis. Patients with acute illnesses did not attend the clinic and thus, are not featured in our analysis. Patients are classed as being smokers if they have ever smoked cigarettes.
HR was categorized as follows. Baseline HR is the HR measured at the patient’s first visit to the clinic. Final HR is the last recorded HR for the patient, who must have attended the clinic for ≥3 visits. Final HR is based on 3 measurements taken at the last visit, and this reflects the patient’s final HR for the last 6 months of clinic follow-up. We measured the difference between baseline HR and final HR to obtain the HR change (ΔHR) for each patient. We analyzed ΔHR as both continuous (ΔHR continuous) and categorical (ΔHR categorical) variables. Finally, comparing baseline and final HR and using an HR threshold of 80 bpm, we classified patients into those who had a consistently high (high-high) or low (low-low) HR, patients whose HR increased over time (low-high), and patients whose HR decreased over time (high-low). These 4 groups, with roughly equal sample sizes, were identified as belonging to 1 of “4 corners” for this analysis (Figure 1). We used 80 bpm as a cutoff in our 4-corner analysis for several reasons. First, it facilitated the division of our cohort into 4 approximately equal groups. Secondly, Palatini,15 using mixture analysis, showed that, within the general population, 2 subpopulations with normal and high HR can be separated at a threshold level of ≈80 to 85 bpm.
Because some antihypertensive agents are HR limiting (β-blockers and nondihydropyridine calcium channel blockers), patients who had been prescribed rate-limiting therapy and who continued with this therapy for ≥1 year during the period of observation were defined as being on rate-limiting therapy. Patients taking rate-limiting therapy for >1 year continued to take rate-limiting therapy until their final visit, and thus, this category reflects the patient treatment at the final HR measurement.
Records kept by the General Register Office for Scotland ensured notification of a subject’s death (provided that it occurred in the United Kingdom) together with the cause of death according to the International Classification of Diseases, 9th Revision, Clinical Modification, 6th ed (ICD-9), codes. We considered CV deaths to be ischemic heart disease (IHD; ICD-9 codes 410 to 414) and vascular deaths composed of deaths from stroke (ICD-9 430 to 438) or other vascular causes (ICD-9 390 to 459). Mortality data were censored at 20 years of follow-up for each participant.
Continuous variables are shown as means (SDs) and were compared between groups using 1-way ANOVA. Categorical variables are shown as counts and percentages and were compared between groups using χ2 tests for association.
The Cox proportional hazards models were used to determine the risk of all-cause, CV, or IHD mortality for each HR group. The models were adjusted for major CV risk factors: age, sex, body mass index, smoking, rate-limiting therapy, systolic blood pressure (at the time of HR measurement), and cholesterol (in millimoles per liter). The effect of rate-limiting antihypertensive therapy was analyzed in each model. The proportionality assumptions were initially met by analyzing the log minus log plots, and then Schoenfeld residuals were examined for any evidence of deviation from the proportional hazards assumption.16 The proportional hazards assumption was not violated for change in HR univariately or multivariately, including time-dependent covariates for all-cause mortality or CV mortality. All of the analyses were carried out using SPSS version 13.0 for Windows (SPSS Inc) and Stata version 10 (Stata Corp, LP).
We studied 4065 hypertensive patients at baseline (47% men and 22% smokers). The mean duration of clinic follow-up was 897 days (7 to 7087 days). The percentages of patients on rate-limiting therapy at baseline and at last visit were 55% and 53%, respectively. The HR range was 42 to 138 bpm (77±15) at baseline and 40 to 132 bpm (74±13) at the final clinic visit. The HR change from baseline to final visit was −73 to +65 bpm (−3±14). Baseline and final HR were categorized as follows: ≤60, 61 to 70, 71 to 80, 81 to 90, and ≥91 bpm. The demographics of the population by HR category are presented in Tables 1 through 3⇓⇓.
HR and Blood Pressure
At baseline, the relationship between baseline systolic BP and HR was J shaped, with the lowest systolic BP seen in the 61- to 70-bpm HR category. However, at final visit, the relationship between HR and systolic BP was linear. Patients whose HR remained >80 bpm at final visit irrespective of baseline HR had higher systolic BPs than those whose final HR was <80 bpm (Tables 1 through 3⇑⇑).
Cox proportional hazards models showed that patients with an HR of ≤60 bpm had the lowest risk for all-cause and IHD mortality. The lowest risk of a CV event was associated with an HR of 61 to 70 bpm. Relative to the baseline HR category (≤60 bpm), patients with an HR of 81 to 90 bpm had the highest risk of all-cause mortality, with a 39% increase in risk (1.39 [95% CI: 0.95 to 2.04]; P=0.092), as well as the highest risk of death because of CV disease and IHD (Table 4⇓). Adjusted hazard curves for all-cause mortality are shown in Figure 2A.
A total of 3364 patients met the criteria for analysis as part of the final HR group (ie, attended the clinic on ≥3 occasions). Analysis of final HR data showed that the highest risk of all-cause mortality is associated with an HR of 81 to 90 bpm, and the lowest risk is associated with an HR of 61 to 70 bpm. Patients in the 81- to 90-bpm HR category had a 64% increase in risk of an event relative to the lowest HR category (1.64 [95% CI: 1.06 to 2.54]; P=0.026). CV risk was highest in the 81- to 90-bpm category, which had a 75% increase in risk compared with the baseline HR category (1.75 [95% CI: 0.97 to 3.17]; P=0.066). The risk of death because of IHD was highest in the 81- to 90-bpm category and lowest in the 61- to 70-bpm category (Table 4⇑). Adjusted hazard curves for all-cause mortality are shown in Figure 2B.
A total of 3364 patients with baseline visit and final visit HR data were available for analysis. Of these patients, 1797 (53.4%) were on rate-limiting therapy, and of the patients on rate-limiting therapy, 1572 (87.5%) were on β-blockers. Patients with consistently high HRs had a 78% increase in the risk of an event (all-cause) compared with those with consistently low HRs (1.78 [95% CI: 1.31 to 2.41]; P<0.001). Risk was then highest in those with low HRs that became high, followed by those with initially high HRs that decreased over time. The results followed the same pattern for CV mortality: patients in the high-high group demonstrated a 92% increase in risk compared with those in the low-low group (1.92 [95% CI: 1.24 to 2.99]; P=0.004). In contrast with other mortality categories, the risk of IHD mortality was highest in the group of patients whose HR increased over time (low-high; 1.98 [95% CI: 0.95 to 4.14]; P=0.07), although risk was similarly high (1.94 [95% CI: 1.05 to 3.59]) and statistically significant (P=0.035) only in the high-high group. In keeping with all-cause and CV mortality, IHD risk was lowest in patients who had a consistently low HR (low-low; Table 4⇑). Adjusted hazard curves for all-cause mortality are shown in Figure 2C.
A total of 3364 patients were analyzed as a part of our ΔHR (categorical) group (Table 4⇑ and Figure 2D). The highest risk of an all-cause event was associated with patients who had increased their HR by ≥5 bpm at the end of follow-up (1.51 [95% CI: 1.03 to 2.20]; P=0.035). Patients who had decreased their HR by between 1 bpm and 9 bpm had the lowest risk of all-cause mortality. For both CV and IHD mortality, the lowest risk of an event was associated with patients who had decreased their HR by >10 bpm at the end of follow-up. As with all-cause mortality, patients who had increased their HR by ≥5 bpm had the highest risk of a CV event, whereas patients whose HR had not changed or increased by 4 bpm at the end of their time at the clinic had the highest risk of an IHD event.
We analyzed 3364 patients in a Cox proportional hazards model using ΔHR as a continuous variable (Table 4⇑). The results showed that, for each beat ΔHR, there was a 1% change in risk for all-cause, CV, and IHD mortality, which was statistically significant.
The majority of our patients on rate-limiting therapy were taking β-blockers, and the most common β-blocker prescribed in this clinic is atenolol (Tables 1 through 3⇑⇑). The proportion of patients on rate-limiting therapy decreased at final assessment from baseline in the 3 highest HR categories (71 to 80 bpm, 81 to 90 bpm, and ≥91 bpm; Tables 1 through 3⇑⇑). It is likely that many of the patients taking rate-limiting therapy at baseline will have successfully lowered their HR, consequently dropping into a lower HR category and, thus, accounting for this fall in numbers. The effect of rate-limiting therapy was assessed within the models. After adjusting for conventional covariates and HR categories, the use of rate-limiting therapy did not show any independent effect on mortality (Table 4⇑).
Epidemiological studies in the general population have consistently shown that a higher HR is associated with increased risk of CV and non-CV events,17,18 with increased sympathetic activity as a proposed mechanism.19 There is convincing evidence from studies in IHD and heart failure of the benefit of HR reduction on outcomes, primarily with the use of β-blockers and nondihydropyridine calcium channel blockers.20,21 However, in hypertension the relationship is still unclear.
In this study we investigated the relationship between HR and outcomes in a large cohort of clinic hypertensive patients representing a cross-sectional sample of patients with mild-to-severe hypertension. We found that HR was an independent predictor of all-cause, CV, and IHD mortality in these patients. Although this is in keeping with results from other studies,7–11 we showed that the change in HR achieved during follow-up of hypertensive patients was a better predictor of risk than baseline or final HR. We also showed that, after correction for rate-limiting therapy (β-blockers and nondihydropyridine calcium channel blockers), HR remained a significant independent risk factor.
As in INVEST,11 there appeared to be a continuous relationship between baseline HR and adverse outcomes with only weak evidence of a noncontinuous relationship for CV mortality. Again, in keeping with INVEST, on-treatment HR was more predictive than baseline HR. However, unlike the findings from INVEST, we found little indication of a J-shaped relationship for final HR or change in HR. This may be explained by differences in the patient populations studied. We studied a general hypertension cohort, whereas INVEST enrolled participants with coronary artery disease, and the J-shaped relationship was more marked in those with previous myocardial infarction or diabetes mellitus, suggesting that excessive HR reduction may be harmful in hypertension with coronary artery disease and important comorbidities.
We did not find any significant independent association of baseline HR and outcomes, suggesting that this is a less stable trait, possibly influenced by extraneous factors, such as stress and sympathetic activity. There was evidence of regression to the mean when comparing baseline HR with subsequent clinic HRs, indicating that the baseline HR is more variable. The final HR or the HR achieved during follow-up is a more stable trait and is closely correlated with the change in HR from baseline. The analysis of the final HR showed that a HR of >81 bpm is associated with a higher risk, and we believe that this is a more valuable outcome measure.
The use of rate-limiting therapy was not significant in the multivariate analysis, suggesting that the relation between HR and mortality is not related to rate-lowering interventions. Bangalore et al13 showed recently that bradycardia induced by β-blockers (predominantly atenolol, 78% of patients) did not improve CV outcomes in hypertensive patients. In this study, we did not differentiate between the different classes of antihypertensive treatment that could lower HR. Our analysis of the final HR and change in HR reflects HR reduction achieved during hypertension treatment. Although there is a debate about whether HR reduction in hypertension may have deleterious hemodynamic effects by decreasing wave reflection from the periphery, this is not consistent across all agents. Among the β-blockers, atenolol has been associated with differential effects on peripheral and central pressure14 compared with vasodilating β-blockers.22 The finding from this study questions whether such hemodynamic differences are of major clinical relevance in routine clinical practice.
When we analyzed outcomes with conventional risk factors and the use of rate-limiting therapy, but excluding the achieved HR in the model, we found that patients on rate-limiting therapy had a poorer outcome, but when HR was included in the model, this relationship disappeared. One explanation could be the increase in central pressure associated with the decrease in HR, but it is also possible that nondihydropyridine calcium channel blockers and β-blockers are preferentially prescribed to people with underlying IHD. Thus, hypertensive patients on rate-limiting treatment may be at a higher CV risk because of preexisting comorbidities. When HR was included in the model, the detrimental effect of rate-limiting therapy disappeared, and a more linear relationship between decreasing HR and better outcomes was observed. This would indicate that the detrimental effect seen with HR lowering therapy may not be because of HR reduction and consequent increase in central pressure but more likely because of higher baseline risk.
The change in HR during treatment showed that those whose HR decreased during the follow-up period had better outcomes. Although this trait is correlated with the final HR, the change in HR addresses additional information, such as treatment-induced HR reduction, blood pressure control, and control of sympathetic overactivity. The HR categories may also reflect blood pressure control as we show that individuals with higher final HR had significantly higher systolic and diastolic blood pressures. There was no linear trend between increasing HR categories and creatinine, glucose, cholesterol, or smoking prevalence. We also note that patients whose HR increased or remained high throughout follow-up had significantly higher blood pressures. Thus, HR in treated hypertensive patients may be a surrogate for blood pressure control; however, our multivariate model included the achieved systolic blood pressure, indicating an additional effect of HR. This observation is consistent with results from the Hypertension and Ambulatory Recording Venetia Study, where baseline clinic HR and HR changes during follow-up were independent predictors for the development of sustained hypertension in young people with stage 1 hypertension.23
The procedures for HR recording in this study complied with European Society of Hypertension consensus recommendations.24 Patients rested for 5 minutes before measurement of pulse rate over 1 minute, and an average of 3 measurements was taken. Readings were made by nurses when HR was expected to be lower than when measured by doctors. Although electrocardiographic recording is more precise, this provides no practical advantage. Thus, our findings are likely to be reliable.
The main limitations of the study are that it was observational and it was derived from a blood pressure clinic population. We do not have data on the physical activity of patients to facilitate its inclusion in our proportional hazards model. However, this population is more representative than clinical trial cohorts, and the findings add to the existing knowledge of HR and long-term outcomes. Additional studies should include the analysis of different antihypertensive drug classes and drug dosages, the effect of comorbidities, measurement of central pressure, and sympathetic activity.
Resting HR in treated hypertensive patients is a marker of risk, and patients whose HR does not reduce during treatment/follow-up should be targeted for more intensive blood pressure control strategies. Whether HR reduction during treatment is merely a marker of improved CV status, whether rate-limiting therapy might after all provide added value in hypertension, or whether both are true remains unsettled.
Sources of Funding
S.P. was supported by an intermediate research fellowship from the British Heart Foundation (FS/05/095/19937) until April 2009. Servier is acknowledged for partial support, which helped complete data collection.
S.P. has received speaking honoraria from Novartis, Servier, and Sanofi Aventis.
- Received October 7, 2009.
- Revision received October 24, 2009.
- Accepted December 2, 2009.
Diaz A, Bourassa MG, Guertin MC, Tardif JC. Long-term prognostic value of resting heart rate in patients with suspected or proven coronary artery disease. Eur Heart J. 2005; 26: 967–974.
Benetos A, Rudnichi A, Thomas F, Safar M, Guize L. Influence of heart rate on mortality in a French population: role of age, gender, and blood pressure. Hypertension. 1999; 33: 44–52.
Thomas F, Rudnichi A, Bacri AM, Bean K, Guize L, Benetos A. Cardiovascular mortality in hypertensive men according to presence of associated risk factors. Hypertension. 2001; 37: 1256–1261.
Lechat P, Hulot JS, Escolano S, Mallet A, Leizorovicz A, Werhlen-Grandjean M, Pochmalicki G, Dargie H. Heart rate and cardiac rhythm relationships with bisoprolol benefit in chronic heart failure in CIBIS II Trial. Circulation. 2001; 103: 1428–1433.
Kolloch R, Legler UF, Champion A, Cooper-Dehoff RM, Handberg E, Zhou Q, Pepine CJ. Impact of resting heart rate on outcomes in hypertensive patients with coronary artery disease: findings from the International Verapamil-SR/Trandolapril Study (INVEST). Eur Heart J. 2008; 29: 1327–1334.
Palatini P, Thijs L, Staessen JA, Fagard RH, Bulpitt CJ, Clement DL, de Leeuw PW, Jaaskivi M, Leonetti G, Nachev C, O'Brien ET, Parati G, Rodicio JL, Roman E, Sarti C, Tuomilehto J. Predictive value of clinic and ambulatory heart rate for mortality in elderly subjects with systolic hypertension. Arch Intern Med. 2002; 162: 2313–2321.
Williams B, Lacy PS, Thom SM, Cruickshank K, Stanton A, Collier D, Hughes AD, Thurston H, O'Rourke M. Differential impact of blood pressure-lowering drugs on central aortic pressure and clinical outcomes: principal results of the Conduit Artery Function Evaluation (CAFE) Study. Circulation. 2006; 113: 1213–1225.
Palatini P. Need for a revision of the normal limits of resting heart rate. Hypertension. 1999; 33: 622–625.
Grambsch PM, Therneau TM. Proportional hazards tests and diagnostics based on weighted residuals. Biometrika. 1994; 81: 515–526.
Kjekshus J, Gullestad L. Heart rate as a therapeutic target in heart failure. Eur Heart J. 1999; 1: H64–H69.
van der Vring JA, Daniels MC, Holwerda NJ, Withagen PJ, Schelling A, Cleophas TJ, Hendriks MG. Combination of calcium channel blockers and beta blockers for patients with exercise-induced angina pectoris: a double-blind parallel-group comparison of different classes of calcium channel blockers-the Netherlands Working Group on Cardiovasc Research (WCN). Angiology. 1999; 50: 447–454.
Palatini P, Benetos A, Grassi G, Julius S, Kjeldsen SE, Mancia G, Narkiewicz K, Parati G, Pessina AC, Ruilope LM, Zanchetti A. Identification and management of the hypertensive patient with elevated heart rate: statement of a European Society of Hypertension Consensus Meeting. J Hypertens. 2006; 24: 603–610.