Authors: Xuefei Hou, Suru Yue, Zihan Xu, Xiaolin Li, Yingbai Wang, Jia Wang, Xiaoming Chen, Jiayuan Wu
Categories: Original Article, hypertension, joint effect, risk factor, stroke
Source: The Journal of Clinical Hypertension
Doi: 10.1111/jch.14905
Recent guidelines have recognized several factors, including blood pressure (BP), body mass index (BMI), low‐density lipoprotein cholesterol (LDL‐C), hemoglobin A1c (HbA1c), smoking, and physical activity, as key contributors to stroke risk. However, the impact of simultaneous management of these risk factors on stroke susceptibility in individuals with hypertension remains ambiguous. This study involved 238 388 participants from the UK Biobank, followed up from their recruitment date until April 1, 2023. Cox proportional hazard models with hazard ratios (HRs) and 95% confidence intervals (CIs) were used to illustrate the correlation between the joint modifiable risk factor control and the stroke risk. As the degree of risk factor control increased, a gradual reduction in stroke risk was observed. Hypertensive patients who had the optimal risk factor control (≥5 risk factor controls) had a 14.6% lower stroke risk than those who controlled 2 or fewer (HR: 0.854; 95% CI: 804–0.908; p < 0.001). The excess risk of stroke linked to hypertension slowly diminished as the number of controlled risk factors increased. However, the risk was still 25.1% higher for hypertensive patients with optimal risk factor control as compared to the non‐hypertensive population (HR: 1.251; 95% CI: 1.100–1.422; p < 0.001). The protective effect of joint risk factor control against the stroke risk due to hypertension was stronger in medicated hypertensive patients than in those not medicated. This finding leads to the conclusion that joint risk factor control combined with pharmacological treatment could potentially eliminate the excess risk of stroke associated with hypertension.
Keywords: hypertension, joint effect, risk factor, stroke
In 2020, stroke was globally ranked second as a cause of death and third for resulting disability, marking a significant rise over the past 30 years [1]. The incidence of strokes globally rose by 70% from 1990, with prevalence increasing by 85%, mortality by 43%, and disability‐adjusted life years (DALYs) due to stroke rising by 32% in 2019 [1]. Currently, around 25 million new stroke cases, 13 million deaths, and 300 million DALYs are recorded annually worldwide [2]. This surge in global stroke burden is not solely attributed to population growth and ageing but also the marked increase in exposure to critical risk factors, such as hypertension, diabetes, adiposity, dyslipidemia, smoking, insufficient physical activity, and kidney dysfunction, emphasizing the inadequacy of present primary prevention measures [3].
Hypertension is a complicated disorder that affects several organ systems. It is the most important modifiable risk factor for stroke [4]. Over 50% of stroke incidents across all age groups can be linked to high blood pressure (BP). Research suggests that a 20 mmHg rise in systolic blood pressure (SBP) corresponds to a 35% higher risk for ischemic stroke, 44% for intracerebral hemorrhage, and 43% for subarachnoid hemorrhage [5]. Moreover, stroke mortality doubles with every 20 mmHg increase in SBP or a 10 mmHg increase in diastolic blood pressure (DBP) [6]. It is also worth noting that rising BP might be a surrogate marker of other stroke risk factors, such as weight gain, reduced glucose tolerance, and metabolic syndrome [7]. As such, stroke prevention in individuals with hypertension is crucial to lessen the burden of stroke and improve these patients’ quality of life.
Around 90% of stroke cases are attributable to modifiable risk factors [8]. Timely identification and management of these risk factors are critical for stroke prevention. Guidelines from key organizations like the American Heart Association (AHA), the European Society of Cardiology (ESC), and the European Stroke Organization underline that primary prevention of stroke should involve early detection and control of main causal and modifiable risk factors, including hypertension, dyslipidemia, diabetes, cigarette smoking, and obesity, to prevent a first stroke event [9, 10]. Multiple credible studies endorse a variety of interventions addressing individual risk factors, such as controlling BP, and cholesterol levels and quitting smoking, to prevent or postpone hypertension complications [11, 12, 13]. Since the risk of stroke notably increases in individuals with multiple risk factors, joint risk factor control may be more beneficial than handling them individually. Nonetheless, it remains unclear to what extent risk factor control could reduce or eliminate the excess risk of stroke in hypertensive patients.
We examined the association between the degree of joint risk factor control and stroke risk in hypertensive patients using a prospective cohort from the UK Biobank study. Moreover, we juxtaposed hypertensive individuals with their non‐hypertensive counterpart from the same cohort to gauge how the excess stroke risk tied to hypertension could be lessened or eradicated.
This study, approved by the UK Biobank (ID: 97101), enrolled participants with hypertension at baseline (more details about data sources are in Supplemental File). We identified hypertensive patients based on self‐reported history, physician diagnosis, medication history, and inpatient diagnosis records. Of the 134 541 individuals with hypertension initially identified, we excluded 12 960 patients with pre‐existing cardiovascular diseases (CVDs [stroke, atherosclerosis, atrial fibrillation, heart failure, and myocardial infarction]) and 40 728 cases lacking complete information on risk factors. To manage confounding factors, we used propensity score matching to pair hypertensive and non‐hypertensive subjects based on age and sex at a 2 ratio. Given that this study was a hypothesis‐based epidemiological study, no attempt was made to estimate the sample size. Instead, all eligible participants in the UK Biobank database were included to achieve maximum statistical power.
According to the 2021 guideline from the ESC and prior literature, we examined six key modifiable risk factors in our study, including BP, body mass index (BMI), hemoglobin A1c (HbA1c), low‐density lipoprotein cholesterol (LDL‐C), smoking, and physical activity [14, 15, 16, 17, 18]. At baseline, the participants of the UK Biobank study had their BP measured in a quiet state using the Omron HEM‐7015IT digital BP monitor, a second measurement was performed after 1 min of rest, and the average of the two measurements was taken as the baseline BP value. According to the guidelines of the ESC/European Society of Hypertension, hypertension was defined as SBP ≥140 mmHg and/or DBP ≥90 mmHg [15, 19, 20]. BMI control was identified as a normal BMI ranging from 18.5 to 24.9 kg/m^2^, following World Health Organization standards [19]. HbA1c control meant less than 48 mmol/mol [21], and LDL‐C was under control at below 3.6 mmol/L [22]. Non‐current smokers were classified as having smoking under control. Physical activity was considered under control with over 150 min of moderate‐intensity or over 75 min of high‐intensity exercise weekly [17, 23]. Full details of risk factor control are shown in Table S1.
This study included the following age, sex, race, Townsend deprivation index (TDI, with negative values indicating high socioeconomic status), alcohol consumption (categorized as never, past, and current), family history of stroke, sleep duration, diet pattern, diabetes, high‐density lipoprotein cholesterol (HDL‐C), and usage of lipid‐lowering, anti‐hypertensive, and hypoglycemic drugs. These covariates were gathered upon enrollment. For variables collected more than twice, the mean values were used. In terms of healthy diet patterns, we considered a healthy diet score greater than or equal to 4, based on the Mediterranean Diet and Heart‐healthy Dietary Priorities recommendations for chronic disease risk reduction [24, 25, 26]. Full details of the covariates are listed in Table S2. Methods for handling missing data are in Table S3.
Participants were followed up from the date of recruitment until an event such as incident stroke, death, loss to follow‐up, or the end of the follow‐up period on April 1, 2023, with whichever event came first. Stroke Incidents were detected using defined algorithms in the UK Biobank, which incorporated information from baseline assessments with relevant data from hospital records and death registries. Incident stroke was defined according to the International Classification of Diseases 10 diagnosis codes, including I60, I61, I62.9, I63, I64, I67.8, I69.0, and I69.3.
We initially categorized hypertensive patients based on their levels of risk factor control at the outset. However, due to some groups having limited participants, patients with Control 2 or fewer risk factors were consolidated into one group, as were those with Control 5 or more risk factors. We utilized Poisson regression to calculate crude incidence rates based on the number of risk factor controls in hypertensive and paired non‐hypertensive populations. The results were articulated as the number of events per 1000 person‐years.
Continuous variables were described as means and standard deviations, or median and interquartile ranges (IQRs). Categorical variables were presented as frequencies and percentages. Cox proportional hazards models were conducted to assess the association between risk factor control level and stroke risk with hypertensive patients having ≤2 risk factors control as reference. The proportional hazards hypothesis was tested using Schoenfeld residual tests, and no violations were observed. Five different models were formulated to account for potential confounding factors. Details about the adjustment of variables in each model are in Supplemental files. The number of risk factor controls was also modeled as a continuous variable with similar covariates. Restricted cubic spine plots based on the Cox regression model were performed to check for any potential non‐linear relationships between risk factor control and stroke risk. Furthermore, Cox regression models were also utilized to compare the excess risk stroke in hypertensive patients with matched non‐hypertensive people concerning the degree of risk factor control. Lastly, the cumulative stroke risk for hypertensive patients with different levels of risk factor control was graphed using the Kaplan‐Meier method.
In subgroup analysis, Cox proportional hazards models were employed for stratified analyses by sex, age, family history of stroke, anti‐hypertensive drug usage, diabetes drug usage, and lipid‐lowering drug usage. These analyses compared hypertensive patients with matched non‐hypertensive subjects while adjusting for the same variables. Additionally, two sensitivity analyses were conducted to assess the robustness of the results. Details about sensitivity analyses are in the Supplemental files. All analyses were performed using R software version 4.3.2. A two‐sided p‐value of less than 0.05 was considered statistically significant.
The flow chart of participant selection in this study is shown in Figure 1. A total of 238 388 participants, including 80 853 hypertensive patients and 157 535 matched non‐hypertensive individuals, were included in this study.
FIGURE 1 Participant selection. BMI indicates body mass index; DBP, diastolic blood pressure; HbA1c, Hemoglobin A1c; LDL‐C, low‐density lipoprotein cholesterol; SBP, systolic blood pressure.
Table 1 presents the baseline characteristics of both hypertensive patients and their matched non‐hypertensive subjects. Of all hypertensive patients, 21 608 (26.72%) had two or fewer risk factor controls, while 34 311 (42.44%), 19 810 (24.50%), and 5124 (6.34%), 3, 4, and ≥5 had control 3, 4, and 5 or more risk factors, respectively. Individuals with a higher degree of risk factor control were more likely to have a higher socioeconomic status. They were also more likely to have never smoked or consumed alcohol, engaged in healthy physical activity, followed a healthy diet, and showed lower SBP, DBP, BMI, and HbA1c values. Except for BP, the highest control rate was observed in HbA1c, followed by LDL‐C and smoking among both the hypertensive and non‐hypertensive populations. Summary of risk factor control rates among the hypertensive and non‐hypertensive populations are shown in Table S4.
A total of 5128 stroke cases were identified among hypertensive patients over a median follow‐up of 13.9 years. The link between the amount of risk factor control and stroke risk appeared to be non‐linear (p for non‐linearity = 0.042). Even after adjustments for various variables, the stroke risk in hypertensive patients decreased significantly with the increase in the control of risk factors (Figure 2A). Females, despite controlling the same number of risk factors, exhibited a higher stroke risk than males (Figure 2B). As detailed in Table 2, each additional controlled risk factor was associated with a 3.3% decrease in stroke risk (HR, 0.967; 95% CI, 0.940–0.994; p < 0.001). Optimal control was established at more than five risk factors, offering an HR of 0.854 (95% CI, 0.804–0.908; p < 0.001), considerably lower than those managing two or fewer risk factors in the fully adjusted model. Both competing risk model‐based sensitivity analyses and those omitting participants with missing data yielded consistent trends (Tables S5 and S6).
FIGURE 2 Association of the degree of risk factor control with stroke risk in hypertensive individuals.(A) Adjusted for age, race, sex, TDI, family history of stroke, healthy diet, sleep duration, drinking, diabetes, HDL‐C, lipid‐lowering drug usage, antihypertensive drug usage, and hypoglycemic drug usage.(B) Adjusted for age, race, TDI, family history of stroke, healthy diet, sleep duration, drinking, diabetes, HDL‐C, lipid‐lowering drug usage, antihypertensive drug usage, and hypoglycemic drug usage. CI indicates confidence interval; HDL‐C, high‐density lipoprotein cholesterol; HR, hazard ratio; TDI, Townsend deprivation index.
During the follow‐up period, stroke incidence was reported in 5128 hypertensive patients and 6260 matched non‐hypertensive participants. As shown in Table 3, the incidence rate of stroke was higher among hypertensive patients, regardless of the level of risk factor control. Furthermore, the incidence rate of stroke among hypertensive patients decreased with increased risk factor control. The highest incidence rate was noted in hypertensive patients managing two or fewer risk factors, with a rate of 6.512 per 1000 person‐years (95% CI, 5.039–7.033; Figure 3).
FIGURE 3 Cumulative incidence of stroke stratified by the levels of joint risk factor control in hypertensive individuals.
Similarly, the stroke risk for hypertensive patients was significantly higher compared to that of the matched non‐hypertensive subjects, regardless of the degree of risk factor control. The highest risk was observed in hypertensive patients with ≤2 risk factor controls (HR, 1.344; 95% CI, 1.250–1.444; p < 0.001) in the fully adjusted model. Moreover, even hypertensive patients with optimal risk factor control (≥5 risk factor control) had a 25.1% excess risk of stroke compared to matched non‐hypertensive subjects (Table 3). These findings remained consistent after adjusting for competing events such as death before stroke or excluding participants with missing data, as demonstrated in Tables S5 and S6.
Figure 4 illustrates the results of subgroup analyses based on parameters like sex, age, family history of stroke, and the use of drugs for hypertension, hypoglycemia, and lipid‐lowering. In each subgroup, hypertensive patients showed a higher stroke risk than their non‐hypertensive counterparts, irrespective of the levels of risk factor control. This risk decreased as control improved. Similar results were also observed in subgroup analyses by LDL‐C, BP, and glucose management (Figure S1). However, it is worth noting that hypertensive patients on any of the mentioned medications who effectively controlled more than three risk factor controls did not present significant excess stroke compared to non‐hypertensive participants (p < 0.05). This was not true for hypertensive patients not on medication, who had a markedly higher stroke risk regardless of their degree of risk factor control.
FIGURE 4 Subgroup analyses for stroke risk in hypertensive individuals with different levels of risk factor controls compared to matched non‐hypertensive participants. CI indicates confidence interval; HR, hazard ratio.
In this prospective cohort study, we discovered that increasing the number of modifiable risk factor controls could reduce stroke risk in hypertensive patients. Specifically, each additional risk factor control was linked with a 3.3% decrease in stroke risk. Hypertensive patients with optimal risk factor management (five or more controls) had a 14.6% lower stroke risk than those with two or fewer controls. Although the excess stroke risk due to hypertension steadily decreased as more risk factors were controlled, it was still 25.1% higher in optimally managed hypertensive patients compared to the non‐hypertensive population. It was important to note that the joint management of risk factors had a stronger protective effect against excess stroke risk in medicated hypertensive patients than in those not taking medication.
Stroke has emerged as a global public health issue, causing a severe financial burden. There is an immediate need for a primary prevention strategy to lessen the stroke burden. According to the World Stroke Organization, one way to achieve this is through integrated prevention approaches to address both the general population and those with an escalated risk [27]. This should involve early detection and management of hypertension. Numerous observational studies have confirmed that hypertensive individuals have a heightened stroke risk and lower life expectancy despite meeting BP control targets. This suggests that merely controlling BP is sufficient; comprehensive control of multiple risk factors is essential to prevent stroke in hypertensive patients [13]. In this study, we discovered that optimally managing risk factors could not negate the excess stroke risk brought on by hypertension entirely. This finding contradicts early research examining the association between joint risk factor control and heart failure risk in hypertensive patients [17]. The higher sensitivity of the brain to high BP than the heart potentially explains this residue risk for cerebrovascular diseases despite meeting BP targets [28]. Furthermore, the average duration of hypertension in this study was 13.9 years. This long exposure might increase stroke risk even with good control and may not be entirely offset by later comprehensive risk factor management.
To lessen the burden of stroke, board surveillance of its risk factors is necessary, particularly in a resource‐limited environment. In these settings, it is crucial to collect data on prevalent risk factors that are not only suitable for individual or population‐level interventions but also easy and cost‐effective to monitor. According to existing guidelines and prior studies, we selected certain modifiable risk factors for this study based on their strong predictive link to stroke. These factors include BMI, HbA1c, BP, LDL‐C, smoking, and physical activity. From a public health standpoint, large‐scale actions targeting these variables are the most cost‐effective approaches to stroke prevention. By addressing these key risk factors, we can simultaneously prevent and postpone substantial disability and deaths from stroke and other non‐communicable diseases.
Guidelines have identified BP, BMI, HbA1c, LDL‐C, smoking, and physical activity as risk factors for stroke [9, 20, 29], but the potential mechanisms and the interactions between these risk factors remain unclear. According to the Global Burden of Disease 2019 study, high BMI was the most rapidly increasing risk factor for stroke between 1990 and 2019 [1]. Adipose tissue produces numerous bioactive compounds, which can cause changes in energy balance, immune response, insulin sensitivity, lipid metabolism, and inflammation. These changes can lead to endothelial dysfunction and atherosclerosis, both primary causes of CVDs, including stroke [28]. Hypertension is most significantly affected by BMI, with both conditions worsening each other [30]. Deliberate weight reduction, especially the targeted loss of fat mass, has been proven effective in lowering both stroke incidence and mortality rates among obese patients [31]. BP management can effectively deter both macrovascular and microvascular complications while controlling LDL‐C levels primarily suppresses macrovascular complications, and blood glucose control aids in preventing microvascular complications [32, 33]. Randomized controlled studies (RCTs) have lent support to the notion of a multifactorial approach targeting HbA1c, BP, and lipid levels to prevent cerebrovascular events, especially stroke, in individuals with type 2 diabetes [34]. As of now, no RCT has unveiled the impact of multifactorial interventions on stroke prevention in hypertensive patients. Nevertheless, our real‐world data‐based studies have revealed that maintaining HbA1c, BP, and LDL‐C levels below target thresholds is associated with a reduced risk of stroke.
The American Stroke Association (ASA) has reported that physical activity “reduces BP, improves endothelial function, reduces insulin resistance, improves lipid metabolism, and may help reduce weight” [29]. Moreover, even minimal smoking habits can increase the risk of stroke; smoking as little as one cigarette per day is associated with a 25%−30% increased stroke risk [35]. A large comprehensive meta‐analysis concluded that even light smoking carried a substantial risk of coronary heart disease and stroke, indicating that no safe level of smoking exists when it comes to CVDs. It is, therefore, crucial for smokers to aim for complete smoking cessation rather than merely reducing their intake to reduce stroke risk [36]. Further, the duration of smoking cessation correlates directly with a lower risk of chronic heart disease and stroke, attributable to a reduction in arterial stiffness, oxidative stress, and inflammation [13]. In conclusion, different risk factors may contribute to stroke risk through various mechanisms. Co‐regulation of these risk factors can help mitigate the exaggerated risk of stroke associated with hypertension.
According to previous reports, only 18%−36.8% of hypertensive people adhered to drug treatment, and less than half of hypertensive patients receiving pharmacological therapy maintained adequate BP control [37, 38, 39, 40]. A significant primary cause of CVDs is unhealthy lifestyle choices such as smoking, excessive alcohol consumption, unhealthy diet habits, and lack of regular physical exercise. The FINGER trial demonstrated that multi‐domain lifestyle intervention can reduce the risk of cerebrovascular events, including stroke. Interestingly, participants who were not on previous anti‐hypertensive medication seemed to benefit more [41].
We found that the risk of stroke due to hypertension is lower in males than in females, even when they have similar levels of risk factor control. Even though previous studies have highlighted a similar link between hypertension and stroke risk for both sexes, the diagnosis, treatment, and control rates of hypertension are substantially different between both sexes. This may be attributed to distinct lifestyle choices, environmental exposures, healthcare disparities, and biological differences between the sexes [42]. Additionally, certain risk factors affect outcomes, such as BP and CVDs, in a sex‐specific way. For example, women are more prone to traditional cardiovascular disease risk factors, including central obesity, high blood lipids, and physical inactivity, while men are more likely to engage in smoke and drinking [6]. Therefore, employing hypertension management strategies tailored to each gender and their risk profile can help mitigate stroke risks and improve health outcomes.
We noted that patients undergoing drug therapy showed a stronger protective effect from controlling joint risk factors, irrespective of the types of medication used. Specifically, controlling more than three risk factors can dramatically reduce the stroke risk associated with hypertension in patients receiving drug treatment. This is because the 95% CI of the HRs covered the value of 1. However, just because a finding is not statistically significant does not mean no heightened risk exists. The HRs for controlling joint risk factors were considerably above 1, probably due to insufficient cases reducing statistical power. Consequently, these findings must be interpreted cautiously. Overall, our findings suggest that hypertensive individuals not taking medication appear less responsive to the control of multiple risk factors. As a result, more proactive prevention strategies should be established to enhance stroke prevention.
This study has several advantages. It is based on a large, prospective cohort of hypertensive patients, with matched control subjects for comparison. The extensive data set from the UK Biobank enabled us to analyze the association between joint risk factor control and stroke risk while considering numerous potential confounders. Consequently, our findings are statistically robust.
Our study also has several limitations. Firstly, this study, aligning with the ACC/AHA guidelines, focused primarily on the most influential modifiable risk factors for stroke in hypertensive patients, encompassing lifestyle and biomarkers. However, this may not be the optimal set of factors. Amplifying our scope to include additional risk factors, such as air pollution, frailty, stress, and socioeconomic status, coupled with a more comprehensive assessment of diet patterns and physical activity, could make for interesting future research. Secondly, although we adjusted for major confounders and performed sensitivity analyses, biases resulting from unknown or unmeasured factors persist. As stroke bears some genetic susceptibility, this variable was not accounted for in our study due to a high number of missing stroke‐related genes data in the UK Biobank database, and the current data does not advocate the use of genomic risk scores in primary prevention of CVDs [9]. Lastly, our study was conducted based on the UK Biobank cohort, primarily composed of white British (>90%) who generally display healthier behaviors than the wider UK population, therefore, the broad applicability of our findings may be restricted.
In this cohort study, we found that controlling joint risk factors can lower stroke risk in hypertensive patients. However, even when all risk factors were optimally managed, a significant risk of stroke due to hypertension remained. Significantly, if pharmacological treatment is utilized for joint risk factor control, it may eliminate excessive stroke risk related to hypertension. These new findings indicate that each additional modifiable risk factor not at optimal level increases the risk of stroke related to hypertension. Therefore, it is unrealistic to isolate the effect of these risk factors on stroke incidence. Instead, aggressive management of multiple risk factors in hypertensive patients can lead to gradual improvement in their health outcomes.
Concept and Jiayuan Wu, Xiaoming Chen, and Xuefei Hou. Acquisition, analysis, or interpretation of Xiaolin Li, Zihan Xu, and Yingbai Wang. Drafting of the Jiayuan Wu, Xuefei Hou, Suru Yue, and Jia Wang. Statistical Xuefei Hou, Zihan Xu, and Yingbai Wang.
The funding sources and sponsor 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.
UK Biobank data have approval from the North West Multi‐Centre Research Ethics Committee (MREC) (REC 21/NW/0157). This research has been conducted with the UK Biobank Resource under project 97101.
The authors declare no conflicts of interest.
Xiaoming Chen, Email: chenxiaoming@gdmu.edu.cn.
Jiayuan Wu, Email: wujiay@gdmu.edu.cn.
Data analyzed in this study was made available through the UK Biobank. Data are available upon application to UK Biobank https://www.ukbiobank.ac.uk.
Data analyzed in this study was made available through the UK Biobank. Data are available upon application to UK Biobank https://www.ukbiobank.ac.uk.