Dose-response association of resting heart rate and hypertension in adults

Supplemental Digital Content is available in the text

Abstract

Background:

The association of resting heart rate (RHR) and hypertension in adults is unclear. We aimed to perform a meta-analysis of cohort studies to clarify the association.

Methods:

We searched PubMed and Embase from their inception to November 3, 2017, for published articles. We used a random effects model to combine study-specific relative risks (RRs) and 95% confidence intervals (CIs). We used restricted cubic spline functions to assess the dose-response relationship.

Results:

Nine cohort articles (12 independent studies) with 79,399 individuals and more than 26,380 incident cases of hypertension were included. The summary RR for hypertension was 1.09 (95% CI: 1.06–1.13) with each 10 bpm increment in RHR. The cubic spline model suggested that when compared with 55.5 beats per minute, the risk of hypertension significantly increased with increasing levels of RHR (P_nonlinearity = 0.059).

Conclusion:

We found a linear dose-response association between RHR and incident hypertension in adults.

1. Introduction

Resting heart rate (RHR), a sensitive, noninvasive and inexpensive indicator of cardiac function, has attracted widespread concern in recent years. Several dose-response meta-analyses ascertained that elevated RHR increased the risk of metabolic syndrome,^[1] type 2 diabetes mellitus,^[2] atrial fibrillation,^[3] cardiovascular disease and all-cause mortality.^[4,5] Hypertension, a major cardiovascular risk factor and a leading cause of global deaths, caused an estimated 9.4 (95% confidence interval [CI]: 8.6–10.1) million deaths and accounted for 7.0% (95% CI: 6.2–7.7) of disability adjusted life years in 2010.^[6]

Cross-sectional or nest case-control studies have demonstrated that elevated RHR was associated with an increased risk of hypertension.^[7,8] Among cohort studies, some studies suggested an increased risk of hypertension with elevated RHR,^[9–13] whereas others found the association could be affected by sex and age.^[14] Despite a dose-response meta-analysis^[15] of RHR and risk of hypertension being published during the writing of the current paper, it missed an eligible cohort study^[16] and included 2 noneligible cohort studies (one^[17] with participants <10 years of age whose normal RHR were higher than adults and the other^[18] about the association of RHR and elevated blood pressure [nonhypertension]). These shortcomings might have introduced a selection bias.

Hence, we conducted a systematic review and meta-analysis of cohort studies that examined the dose-response relationship of RHR and hypertension in adults.

2. Methods

2.1. Literature search strategy

We systematically searched for studies investigating the association of RHR and hypertension published in English until November 3, 2017 in the PubMed and Embase databases by using the terms ((“hypertension” OR “high blood pressure” OR “high blood pressures”) AND (“heart rate” OR “resting pulse”)) in the title/abstract. The reference lists of relevant studies were also searched.

2.2. Study selection

Studies were included if they

• 1.were cohort study designs,
• 2.were performed on participants ≥18 years of age,
• 3.used a definite hypertension diagnostic method,
• 4.excluded participants with hypertension at baseline,
• 5.reported relative risks (RRs), odds ratios or hazard ratios with 95% CIs or data to calculate them, and
• 6.reported a quantitative measure of RHR, the number of cases, and the exposed person-years/participant numbers for the dose-response analysis.

We excluded studies that

• 1.involved cross-sectional study designs,
• 2.were reviews, comments, or letters,
• 3.reported RHR as a dichotomous variable, and
• 4.contained duplicate data.

If multiple articles were published from the same study, we included data from the study with the most informative reporting of RHR levels or the larger sample size. The year of study publication was used as the time variable. Two reviewers (L.S. and Y.W.) independently performed the literature search to identify studies. Discrepancies were resolved by discussion with a third reviewer (X.J.).

2.3. Data extraction

The following data were independently extracted by two reviewers (L.S. and Y.W.) for each included study: first author, place of study, publication time, follow-up time, sample size, age and sex of participants, number of cases and participants, RHR quantity, odds ratio (OR)/hazard ratio (HR)/relative risk (RR) and 95% CI, method of RHR measurement, and covariates.

2.4. Assessing the risk of bias

Two reviewers (L.S. and Y.W.) used the Newcastle-Ottawa Scale critically appraise the included studies.^[19] The scoring system involved 3 columns: selection of study participants (4 items), intergroup comparability (1 item), and measurement of outcome (3 items). The total score was the sum of scores for each item, with >7, 4-7, and <4 representing high, moderate, and low quality, respectively.

2.5. Statistical analysis

Data are presented as RRs and 95% CIs. For cohort studies reporting HRs or ORs for hypertension, we assumed that the HRs and ORs were approximately RRs.^[20] Heterogeneity of studies was estimated by the Q test. We selected the model to combine HRs, ORs, or RRs for hypertension according to the results of the Q test. If the number of cases in each category was missing, these data were inferred on the basis of the number of total cases and the reported effect size. If the exposed person-years or participant numbers were not reported in each category, groups were assumed to be of equal sizes.^[21]

Generalized least squares regression was used to estimate the study specific dose-response association. The generalized least squares regression model estimates the linear dose-response coefficient by taking into account the covariance for each exposure category within each study because they are estimated relative to a common referent RHR exposure category.^[22,23] The DerSimonian and Laird random effects model was used to pool the study-specific dose-response RR estimates.^[24]

First, a linear association was assumed; study-specific RR estimates were calculated per RHR increments of 10 beats per minute (bpm) and then pooled. In addition, we examined possible nonlinear associations by modeling RHR using a restricted cubic spline with 2 knots located at the 33.3rd and 66.7th percentiles of the distribution.^[25]

Only studies reporting risk estimates for at least 3 RHR exposure levels for incident hypertension were included in this analysis. The P value for nonlinearity was calculated by testing the null hypothesis that the coefficient of the second spline was equal to zero.

To explore the sources of heterogeneity in RHR, we performed subgroup analyses by sex, study design, nationality, number of cases, follow-up time, measurement of RHR, and the covariates (smoking, alcohol drinking, physical activity, body mass index (BMI), fasting plasma glucose (FPG), and blood pressure) adjusted for in the analysis. We also performed a sensitivity analysis by excluding 1 study at a time to assess the stability of results and potential sources of heterogeneity.

Publication bias was evaluated by Egger test and Begg plotting. Sensitivity analyses involved removing some low-quality studies. Statistical significance was set at P < .05. Analyses involved the use of Stata 12.1 (Stata Corp, College Station, TX). All analyses were based on previous published studies; thus, no ethical approval, and patient consent were required.

3. Results

3.1. Search results and study characteristics

We identified 9 articles^{[9–12,16,26–29]} (12 independent studies) in PubMed and Embase for the meta-analysis (Fig. 1). In total, the review included 79,399 individuals and more than 26,380 incident cases of hypertension. The details of the included studies are presented in Table 1. In all, 4 articles^[9,10,12,29] did not distinguish between sex and 3 articles^[11,16,28] described stratified analyses by sex. Overall, 1 study^[26] was conducted in Italy, 2^[9,16] in America, 3^[12,27,29] in Japan, and 3^[10,11,28] in China. Analyses of the quality of studies yielded the Newcastle-Ottawa Scale scores presented in Table S1.

Figure 1.

Flow diagram of studies in the meta-analysis.

Table 1.

Summary of the characteristics of cohort studies investigating the association of level of resting heart rate (RHR) and incident hypertension.

3.2. Dose-response association between RHR and incident hypertension

The summary RR for hypertension was 1.09 (95% CI: 1.06–1.13) with each 10 bpm increment in RHR, and there was low substantial heterogeneity (I² = 53.4%; P_{heterogeneity} = 0.015) (Fig. 2).

Figure 2.

Risk of hypertension for per 10 bpm increment in resting heart rate.

We included 8 studies^{[9–12,28,29]} in the nonlinear dose-response analysis; 4 studies^[16,26,27] reporting only continuous risk estimates were excluded because the analysis required at least three categories of RHR. We found no evidence of a nonlinear association between RHR and hypertension (P_nonlinearity = 0.059), so restricted cubic splines were adopted to model the linear dose-response association. The results from the cubic spline model suggested that when compared with 55.5 bpm, the risk of hypertension increased significantly with increasing levels of RHR (Fig. 3). Compared with 55.5 bpm, the risk of hypertension was 1.05 (95% CI, 1.03–1.07) at an RHR level of 60 bpm and 1.58 (95% CI, 1.32–1.88) at an RHR level of 100 bpm with the linear cubic spline model.

Figure 3.

Linear dose–response association analysis of resting heart rate and incident hypertension modeled by using restricted cubic splines.

3.3. Subgroup and sensitivity analyses and publication bias

Results of the subgroup analyses are presented in Table 2. In general, the association was consistent in most analyses. The heterogeneity seemed to be lower in men and women populations (I² = 0.0, 10.6%), with a duration of follow up (years) < 5 and ≥ 5 (I² = 0.0, 14.6%), in American populations (I² = 0.0), in studies with cases of hypertension <500 (I² = 0.0), with electrocardiography measurements of RHR (I² = 0.0), and when the analyses were smoking-unadjusted (I² = 0.0), physical activity-adjusted (I² = 0.0), FPG-adjusted (I² = 0.0), and BMI-adjusted (I² = 0.0). No significant changes in heterogeneity occurred in other subgroup analyses.

Table 2.

Dose-response subgroup analysis of risk of hypertension per 10 bpm increment of RHR.

No individual study had an excessive influence on the pooled effect in the sensitivity analysis. We found no evidence of publication bias by Egger test and Begg plotting (P = .064).

4. Discussion

To our knowledge, this is the first meta-analysis of cohort studies to quantify the dose–response relationship between RHR and incident hypertension in adults. We found a linear, positive association between increments in RHR and risk of incident hypertension in adults, with a 9% increase in RR per 10 bpm increase in RHR. Compared with 55.5 bpm, those who met the guidelines-recommended minimum RHR levels of 60 bpm and maximum RHR levels of 100 bpm had 5% (RR, 1.05; 95% CI, 1.03–1.07) and 58% (RR, 1.58; 95% CI, 1.32–1.88) higher risk of hypertension, respectively.

In contrast to a previous report,^[15] we quantitatively assessed the risk of hypertension by dose-response analyses for people who met the guidelines-recommended minimum and maximum RHR. This study excluded two studies^[17,18] and added one study^[16] containing 458 cases among 2677 participants in our dose-response model. We observed a significant positive association between the RHR and hypertension, which was consistent with the previous meta-analysis. However, a recent study (The Rural Chinese Cohort Study)^[14] did not identify a significant RR (1.02, 95% CI: 0.96–1.10) of hypertension per 10 bpm increase in men. Because of limited number of studies stratified by sex, further research is essential to confirm the possible relationship in men.

Several biological mechanisms could explain the association of RHR and hypertension. High RHR represents an imbalance in central nervous system activity, thereby leading to increased sympathetic and decreased parasympathetic tone. Increased sympathetic tone is also closely associated with high blood pressure or hypertension, which can be one possible mechanism.^[30] High RHR or autonomic imbalance is also associated with subclinical inflammation represented by elevated C-reactive protein levels and leukocyte counts.^[31,32] Therefore, inflammation may play an important role in the association of RHR and hypertension. One possible link between RHR and hypertension is leptin, an adipocytokine produced by adipose tissue that directly increases sympathetic outflow,^[33] thereby increasing blood pressure and heart rate.^[34,35] Elevated RHR might be a marker of chronic stress and anxiety, which may increase the risk of hypertension.^[36] In a dose-response meta-analysis of cohort studies, the risk of hypertension gradually decreased with increasing exercise, which could be related to reduce sympathetic nerve activity, plasma norepinephrine levels, and total peripheral resistance and to improve endothelial function.^[37–39] Because exercise training can also lower RHR, the risk of hypertension could be reduced with regular exercise via reduced RHR; thus, associations between RHR and hypertension may be worthy of further exploration.

5. Strengths and limitations

Our meta-analysis has several strengths. The broad search of relevant studies minimized meta-analysis. Last, the study quality was high and ranged from scores of 8 to 9, with the exception of one study with the lowest score being 6. Combining the evidence from all available studies increased the statistical power to detect an association between elevated RHR and hypertension. We conducted linear dose-response analyses and incorporated the recommended RHR levels to clarify the strength and shape of the dose-response relationship between RHR and hypertension. We also conducted a sensitivity analysis, which demonstrated that no individual study had an excessive influence on the pooled effect. These findings support a possible role for RHR in the development of hypertension.

Some potential limitations of this meta-analysis should be mentioned. RHR could be influenced by numerous factors.^[40] However, most of the included studies adjusted for the main influencing factors in the multivariable regression models. Therefore, the effect of additional factors might be relatively limited. Heterogeneity among the included studies is unavoidable in all meta-analyses. To discover potential sources of heterogeneity, we performed various subgroup analyses and found that the association did not vary, suggesting that the finding is robust for hypertension.

5.1. Future directions

There is no cutoff at which harms are not achieved and more risks occur with increasing RHR. Future studies examining multiple aspects of RHR are needed to explore the optimal value of RHR for hypertension prevention.

6. Conclusions

This meta-analysis found an increased risk of hypertension with increasing RHR in adults. Elevated RHR may be used as an additional clinical indicator of hypertension in adults.

Author contributions

Conceptualization: Chengyi Han and Yongguang Yang

Data curation: Lijun Shen and Yuming Wang

Formal analysis: Yongcheng Ren

Supervision: Xuesong Jiang

Validation: Xuesong Jiang, Chengyi Han

Writing – original draft: Lijun Shen and Yuming Wang

Writing – review & editing: Chengyi Han and Yongguang Yang