Hypertension is the primary and most common risk factor for heart disease, stroke, and renal disease and has been identified as the leading cause of mortality and third cause of disability-adjusted life years worldwide.^1,2 According to a report from Kearney et al, the total number of adults with hypertension in 2025 was predicted to increase to 1.56 billion worldwide.³ Identifying and characterizing modifiable risk factors of hypertension remain important for public health and clinical medicine.

Genetic and lifestyle risk factors are thought to be associated with hypertension.^4–6 The World Health Organization has developed a series of recommendations based on these factors to prevent and control disease.⁷ Increasing physical activity (PA) is one of these recommendations because it is considered a widely accessible, inexpensive, and effective intervention. PA can be recreational PA (RPA), performed during free time and to meet personal interests and needs, and occupational PA (OPA), associated with the activity required for one’s job.⁸ To our knowledge, the international recommendations for health-promoting PA do not distinguish between RPA and OPA.⁹

Numerous investigations and research have investigated the effect of PA in reducing hypertension risk; however, they have reported conflicting results.^10–15 Li et al¹⁶ performed a meta-analysis to investigate the relationship between PA and risk of cardiovascular disease; however, there has not yet been a meta-analysis to explore the effect of PA in reducing hypertension risk or the differences in how RPA and OPA can decrease the risk of hypertension. Thus, we performed this meta-analysis to investigate the association between PA and incidence of hypertension.

Methods

The Meta-analysis of Observational Studies in Epidemiology (MOOSE) guidelines were followed for the current study.¹⁷

Search Strategy

The PubMed and Embase databases were searched from its inception through November 26, 2012 to identify all relevant literature. The following search strategy was used: (physical activity OR physical activities OR motor activity OR motor activities OR exercise OR exercises OR walking OR energy expenditure) AND (hypertension OR high blood pressure OR high blood pressures) AND (cohort study OR prospective study OR longitudinal study OR follow-up study). The subjects of studies were defined as humans, and the languages of articles were limited to English and Chinese because the reviewers are fluent in both of these languages. Reference lists of all included studies were manually reviewed and those of relevant review articles to identify additional studies.

Inclusion Criteria

To be included in this analysis, a study must have met the following criteria: (1) prospective cohort study; (2) exposure of interest was different levels of RPA, OPA, or commuting PA; (3) outcome of interest was hypertension; (4) study population was healthy without history of hypertension; (5) analysis adjusted for confounding factors; and (6) relative risk (RR) or hazard ratio with 95% confidence interval (CI; or data to calculate them) were reported. If multiple articles were published from the same cohort, the most informative report was included.

Data Extraction and Quality Assessment

The following items were extracted for each study: name of the first author, year of publication, country, duration of follow-up, characteristics of cohort population, number of hypertension cases, type of PA (RPA, OPA, and commuting PA), measurement of PA, adjustment for potential confounding, and estimates of associations. PA was categorized in 3 levels for studies that reported ≥3 levels of PA: the lowest category was defined as low-level PA (reference group), the highest category as high-level PA, all categories in between were pooled to represent moderate-level PA.¹⁸ PA was categorized in high and low level for studies that reported 2 levels of PA. For each selected study, the RR (95% CI) or hazard ratio (95% CI) for the high- versus the low-level PA group and for the moderate versus the low PA group was extracted.¹⁶ We distinguished among RPA, OPA, and commuting PA according to the descriptions in the articles. As described in the studies, RPA is voluntary and purposeful, such as running, jogging, swimming, cycling, ball games, and some other form of exercise, whereas OPA is related to work, such as industrial work, farm work, or forestry work. Commuting PA is related to getting to work and back home, such as walking or cycling. When multiple effect estimates for PA were presented, lifetime PA data were used, otherwise, the most recent PA data were used. Quality assessment for included studies using the Newcastle Ottawa Scale was recommended by the Cochrane Non-Randomized Studies Methods Working Group.¹⁹ Data extraction and quality assessment were performed by 2 independent investigators (P.H. and H.X.). Any disagreement was settled by discussion.

Statistical Analysis

The pooled RR with its corresponding 95% CI was calculated to assess the association of PA with the risk of hypertension. Heterogeneity among studies was assessed using the Q test and the I² statistic.²⁰I² describes percentage of total variation because of between-study heterogeneity rather than chance. In the presence of substantial heterogeneity (I²>50%), the Dersimonian and Laird random effects model (REM) was adopted as the pooling method, otherwise, the inverse-variance fixed effects model (FEM) was applied as the pooling method.²¹ Because several studies only provided RRs of subgroups by sex or race, etc, the overall RR for each study was calculated using REM or FEM based on heterogeneity between subgroups. Meta-regression was conducted to explore the possible sources of between-study heterogeneity. Subgroup analysis by duration of follow-up was performed. A sensitivity analysis was performed to validate the stability of outcomes by sequential removal of each individual study.²² An individual study is suspected to excessively influence the point estimate if its omitted analysis lies outside the 95% CI of the combined analysis. Publication bias was estimated using a funnel plot and modified Egger linear regression test.²³ All statistical analyses were performed with STATA version 11.0 (StataCorp LP, College Station, TX). All tests were 2-sided and a P value <0.05 was considered statistically significant.

Results

Characteristics of Studies

The literature search identified 3817 potentially relevant articles, of which 13 studies ultimately met the inclusion criteria (Figure S1 in the online-only Data Supplement).^{10–15,24–30} The total population of the included studies was 136 846 persons who were initially free from hypertension, and 15 607 persons developed hypertension during follow-up. The follow-up duration ranged from 2 to 45 years, and the median duration of follow-up was 9.8 years. Twelve studies reported the effect of RPA on hypertension risk,^{10–15,24,26–30} 6 reported the effect of OPA,^{11,13,24–26,30} and 2 reported the effect of commuting PA.^27,30 All included studies reported the effect of high-level PA on hypertension risk,^{10–15,24–30} and 9 reported the effect of moderate-level PA.^{11–15,26,27,29,30} Two studies involved men only,^15,27 2 involved women only,^13,29 6 studies involved both men and women and reported sex-specific results,^{14,24–26,28,30} and 3 studies involved both men and women but did not report sex-specific results.^10–12 Seven studies defined hypertension as systolic blood pressure ≥140 mm Hg or diastolic blood pressure ≥90 mm Hg or use of antihypertensive medication,^{10,11,14,15,24–26} 2 studies defined hypertension as systolic blood pressure ≥160 mmHg or diastolic blood pressure ≥95 mmHg or use of antihypertensive medication,^27,30 and 4 studies ascertained hypertension by self-report or from a reimbursement medication registry.^12,13,28,29 Five studies were conducted in North America,^{12,14,15,26,29} 6 in Europe,^{10,11,13,24,28,30} and 2 in Asia.^25,27 Stars in Table S1 indicate the quality of the studies assessed using the Newcastle Ottawa Scale and the maximum score was 9. Five studies were scored 9 stars,^{13,24,26,27,30} 5 studies were scored 8 stars,^{10–12,14,25} and 3 studies were scored 7 stars^15,28,29 in quality assessment. The results of all included studies were adjusted for potential confounding factors (Table S1). Other characteristics of included articles, such as statistic used to estimate RR, blood pressure measurement, and definition of PA levels, were also extracted in Table S2.

RPA and Risk of Hypertension

The association between high-level RPA and risk of hypertension compared with low-level RPA is shown in Figure 1. These was no significant heterogeneity between 12 studies (P_Q=0.171; I²=28.0%), and a FEM was used. The overall result showed that high-level RPA was associated with decreased risk of hypertension compared with the reference group with low-level RPA (RR, 0.81; 95% CI, 0.76–0.85).

• Download figure
• Download PowerPoint

Figure 2 shows the association between moderate-level RPA and risk of hypertension compared with low-level RPA. There was no significant heterogeneity between 9 studies examining moderate-level RPA and risk of hypertension (P_Q=0.613; I²=0.0%), so a FEM was used to pool the RR. The result of the meta-analysis showed that moderate-level RPA decreased the risk of hypertension compared with low-level RPA (RR, 0.89; 95% CI, 0.85–0.94).

• Download figure
• Download PowerPoint

In addition, after exclusion of 3 articles^10,24,28 (Figure 1), which reported only 2 levels of RPA, the pooled estimate of the RR of high- versus low-level RPA and risk of hypertension, based on 9 studies, was also significant (RR, 0.78; 95% CI, 0.72–0.83; Figure S2). The difference of these 2 RRs (RR of high-level RPA and RR of moderate-level RPA based on 9 studies) was significant (P=0.020), and the former RR is smaller than the latter one. Thus, we concluded that there was an inverse dose–response association between RPA and incidence of hypertension.

OPA and Risk of Hypertension

The association between high-level OPA and risk of hypertension compared with low-level OPA is shown in Figure S3. A REM was used because of the heterogeneity between 6 studies (P_Q=0.011; I²=66.3%). The pooled result showed that the association between high-level OPA and risk of hypertension was not statistically significant (RR, 0.93; 95% CI, 0.81–1.08).

Figure S4 shows the association between moderate-level OPA and risk of hypertension compared with low-level OPA. A FEM was used because there was no significant heterogeneity between 4 studies (P_Q=0.454; I²=0.0%). However, the result showed that the association between moderate-level OPA and risk of hypertension was not significant (RR, 0.96; 95% CI, 0.87–1.06).

Commuting PA and Risk of Hypertension

The association between high level of commuting PA and risk of hypertension was not consistent in the 2 included articles. Hayashi et al²⁷ found that high commuting PA decreased the risk of hypertension (RR, 0.71; 95% CI, 0.52–0.97), whereas the association reported by Barengo et al³⁰ was not significant (RR, 0.96; 95% CI, 0.82–1.12). The association between moderate level of commuting PA and risk of hypertension was not significant in 2 studies (Table S2).

Exploration of the Heterogeneity Source

Exploratory univariate meta-regression was performed with the introduction of follow-up duration, study area (North America, Europe, and Asia), publication year, sample size, and number of cases during follow-up. The results of meta-regression indicated that follow-up duration was the main source of heterogeneity both in high-level RPA (meta-regression coefficient, 0.005; 95% CI, 0.001–0.009; P=0.020) and in high-level OPA (meta-regression coefficient, 0.007; 95% CI, −0.001 to 0.014; P=0.065). Thus, the subgroup analyses of the association between RPA/OPA and risk of hypertension by follow-up duration (≥10 years versus <10 years) were conducted. However, the differences between short and long follow-up durations were not significant for either high-level RPA or high-level OPA (Table S3).

Sensitivity Analysis and Publication Bias Evaluation

In the sensitivity analysis, no individual study substantially influenced the pooled RRs for both high- and moderate-level RPA (Figures S5–S8). The shape of the funnel plot to assess publication bias was roughly symmetrical for high- and moderate-level RPA. No publication bias was detected by Egger test for high-level RPA (P=0.052), moderate-level RPA (P=0.301), high-level OPA (P=0.329), or moderate-level OPA (P=0.430).

Discussion

To our knowledge, this meta-analysis represents the first one investigating the association between PA and incidence of hypertension. The current meta-analysis included 13 prospective studies with a total population of 136 846 and 15 607 hypertensive cases found at follow-up. The results of this study suggested that there was an inverse dose–response association between levels of RPA and risk of hypertension.

The results also showed that there was no evidence of an association between high- or moderate-level OPA and developing hypertension. Holtermann et al³¹ hypothesized that OPA increased the risk for long-term sickness absence, which is an acknowledged measure of global health and economic burden in Western societies, whereas RPA is thought to decrease the risk for long-term sickness absence.^32,33 A study conducted among female Filipino workers found that excessive work was associated with poorer health, dissatisfaction with life, poor recuperation from fatigue, and hypertension.³⁴ Generally, high-level OPA consists of heavy lifting, prolonged standing, and highly repetitive work, whereas RPA is often characterized by dynamic contractions of large muscle groups increasing whole-body metabolism and cardiac output with the ability to rest when fatigued.³⁵ The international recommendations for health-promoting PA should distinguish between OPA and RPA. In addition, a meta-analysis of randomized controlled trials supports the blood pressure–lowering potential of dynamic resistance training.³⁶

The mechanism between RPA and hypertension is complex. Generally, exercise reduces blood pressure, systemic vascular resistance, sympathetic activity, plasma renin activity, the homeostasis model assessment insulin resistance index, weight, and abdominal circumference, and that it improves blood lipids.³⁷ First, RPA is helpful in maintaining body weight. A randomized clinical trial conducted among overweight adults suggested that weight loss was effective in lowering systolic and diastolic blood pressures.³⁸ Second, exercise decreased total peripheral resistance. Because mean arterial pressure is determined by cardiac output and total peripheral resistance, reductions in resting cardiac output do not typically occur after chronic exercise, whereas total peripheral resistance will decrease followed by decreased blood pressure.³⁹ A meta-analysis that involved 72 trials also found that aerobic endurance training decreased blood pressure through a reduction of vascular resistance.³⁷ Third, hyperinsulinemia and insulin resistance may contribute to hypertension through the effects of insulin on the retention of sodium, increasing sympathetic nervous system activity, and vascular smooth muscle proliferation.⁴⁰ RPA has been shown to improve insulin sensitivity by Henriksen et al,⁴¹ which is another possible mechanism of antihypertension effect of RPA. The proposed mechanisms include neurohumoral and structural adaptations, but no definitive conclusion has been made on the exact mechanism.³⁹ Furthermore, the indirect association between RPA and decreased risk of hypertension might also be because of the characteristics of physically active subjects, who usually have a healthier life style in general. In other words, they may be younger, smoke less, drink less, have more healthy eating habits, be less stressed, which are protective factors of hypertension.

Previous studies have also found PA and RPA to be associated with other health outcomes besides hypertension. Regular PA and RPA might decrease the risk of lung cancer, prostate cancer, colon cancer, and cardiovascular disease.^16,42–45 Because RPA may reduce the risk of hypertension through weight loss, moderate- to high-level RPA that is sufficient to maintain a normal weight should be recommended and promoted in populations, which may prevent hypertension and other diseases.

This meta-analysis had several strengths. This study included prospective cohort studies to determine the risk of hypertension over time. This meta-analysis also had a large sample size and included studies were adjusted for potential confounding, which increased the accuracy of the effect estimate.¹⁶ However, the potential limitations of this meta-analysis should be considered. First, this meta-analysis only included English and Chinese language articles; eligible articles with other languages were not included in this analysis, which may influence the pooled estimated value. Second, because of the inability to obtain raw data, we could perform only a study-level but not a patient-level meta-analysis, which would have enabled us to adjust for multiple factors. Third, the measurements of PA varied among the 13 included studies with regard to frequency, intensity, and duration, leading to different definition of low-, moderate-, and high-level PA, so the optimal energy that should be consumed to reduce risk of hypertension could not be evaluated.

The results of this meta-analysis suggest that there was an association between levels of RPA and decreased risk of hypertension, whereas there is no significant association between OPA and hypertension.

Perspectives

It is important to calculate the precise energy that one should consume through RPA. In addition, sex, age, race, and obese status should be stratified when the precise energy is calculated. In addition, the association between RPA and decreased risk of hypertension in this meta-analysis might be confounded by various factors. Thus, large-scale randomized, controlled trials are recommended to assess the impact of PA per se on the incidence rates of hypertension.

Footnotes