Association of Blood Pressure With Stroke Risk in Chinese Population: A 6-Year Prospective Cohort Study

Objective: Previous results on the association between blood pressure(BP) and stroke risk were controversial. We investigated the association of BP with stroke risk in China. Method: We included a total of 5,700 adults aged 40-90 years from the China Health and Nutrition Survey (CHNS) cohort. Cox proportional hazards regression models were used to estimate hazard ratios and 95% CIs. Restricted cubic spline analyses were used to explore linear and nonlinear relationships of BP and stroke. Result: With a median follow-up of 6 years, a total of 5,700 individuals were enrolled in our study, of whom 178 developed stroke. Multivariable adjusted Cox models including systolic blood pressure(SBP) and diastolic blood pressure(DBP) showed a strong positive association between SBP and overall stroke. Compared with participants with SBP 0-130 mmHg, the multivariable adjusted HRs[95% CIs] in participants with SBP 130-140, 140-160, 160-180, and ≥ 180 mmHg were 1.08[0.62-1.89], 2.41[1.51-3.86], 2.21[1.16-4.20], and 3.90[1.78-8.55] for overall stroke; 0.65[0.21-2.04], 3.68[1.73-7.83], 2.51[0.84-7.47], and 5.91[1.69-20.60] for ischemic stroke; 1.26[0.50-3.20], 1.19[0.47-3.04], 2.06[0.66-6.41], and 5.10[1.36-19.20] for hemorrhagic stroke. Restricted cubic spline analyses including SBP and DBP showed linear relationships of SBP with overall, ischemic and hemorrhagic stroke. No linear or nonlinear relationships of DBP with overall, ischemic and hemorrhagic stroke were observed. Conclusion:


Introduction
Stroke is the leading cause of death and permanent disability worldwide [1] [2] [3]. In China, the most populous country in the world, the incidence of stroke among adults over the age of 40 years had increased by 8.3% from 2002 to 2013 [4] [5]. Hypertension is the most important modi able risk factor for stroke, with a strong, direct, linear, and continuous relationship between blood pressure(BP) and cardiovascular disease (CVD) risk[6] [7]. Speci cally, a meta-analysis of 61 cohorts recruited between 1950 and 1990 reported that usual blood pressure is strongly and directly related to vascular (and overall) mortality, without any evidence of a threshold down to at least 115/75 mm Hg, in participants aged 40-89 years [7].
In China, the prevalence of hypertension rapidly increased in the past 30 years. However, awareness, treatment and control of hypertension had declined or remained unchanged in China from 2000 to 2010 [5][8], while they had increased signi cantly in the developed countries [9], these may affect the incidence of stroke. Despite the high burden of stroke in China, the association between BP and stroke has not been fully investigated by large prospective studies on Chinese. To better understand the etiology and prevent stroke, more evidence on the association between BP and stroke is needed. Thus, we investigated the association of BP levels with overall stroke and its subtypes among 5,700 adults from nine provinces around China.

Study design
The China Health and Nutrition Survey(CHNS) was conducted by the University of North Carolina at Chapel Hill, and the National Institute for Nutrition and Health at the Chinese Center for Disease Control and Prevention to examine the status of economic, public resources, health and nutrition. All methods were performed in accordance with the relevant guidelines and regulations. Details about the study design were available elsewhere [10] [11]. A multistage, random cluster process was used to draw the samples in the Chinese population. The survey was an ongoing nationwide study that started in 1989 and subsequently conducted in 1991, 1993, 1997, 2000, 2004, 2006, 2009, 2011 and 2015 in nine provinces( Heilongjiang, Liaoning, Shandong, Jiangsu, Henan, Hubei, Hunan, Guangxi, and Guizhou) of China. Baseline data collection included demographic information, medical history, standardized medical examination, laboratory tests, and anthropometric measurements.

Study population
For this study, data were drawn from the 2009-2015 CHNS cycles. We included Chinese residents with BP data collected at age 40 to 90 years and with subsequent follow-up time available(n = 7,717). After excluding persons who were followed up only until 2011 and did not participate in the later follow-up survey in 2015(n = 958) and those without complete BP measure data at baseline(n = 1,059). As a result, 2,680 men and 3,020 women were available for analysis.

Exposure,covariates and outcomes
Adults and children received detailed physical examinations that included weight, height, arm and head circumference, mid-arm skinfold measurements, and BP. BP was measured thrice by experienced physicians with the participant in the sitting position. Systolic blood pressure(SBP) and diastolic blood pressure(DBP) were de ned as mean SBP and DBP of three test results.
History of diseases, individual activities, lifestyle, health status, marriage and birth history were acquired through an individual questionnaire. Smoking status was classi ed into two categories as follows: current smoker or not. Drinking status was divided into two groups: current drinker or not. Diabetes was identi ed by self-reports of a history of diabetes diagnosis. The biomarker data collected in CHNS 2009 involves the release of 26 fasting blood measures on individuals aged 7 and older. Frozen serum samples were sent to a national central lab in Beijing for measurement of serum lipid levels.
Incident stroke was diagnosed by a doctor's diagnosis or treatment history for stroke during the follow-up period (2009)(2010)(2011)(2012)(2013)(2014)(2015). Cases were censored at the date of diagnosis of stroke or the nal visit, whichever came rst.

Statistical analysis
Results are presented as mean ± standard deviation or median (interquartile range) for continuous variables, and number (percentage) for categorical variables. Mantel-Haenszel χ2 test was used for categorical variables. T-test was used for continuous variables which were normally distributed and Wilcoxon rank-sum test was used for variables which were not normally distributed.
We used cox proportional hazards models to explore the relation between BP level and overall stroke incidence [12]. In addition, we have also established models with outcomes of ischemic stroke and hemorrhagic stroke(excluding unknown stroke participants). Fully adjusted models were adjusted for potential confounders including age, gender(male or female), body mass index(BMI, kg/m 2 ), total cholesterol(TC, mmol/L), drinking status(current drinker or not), smoking status (current smoker or not), diabetes(no or yes), CVD history, and site(rural or urban). In addition, to examine the independent association of SBP and DBP with stroke risk, we further ran a multivariable model including both SBP and DBP. Tests for linear trend were computed by modeling the median values of each category as a continuous variable in regression models.
Restricted cubic spline analyses with 4 knots (5th, 35th, 65th and 95th) were also used to explore the potential linear or nonlinear relationship of BP level with overall stroke, ischemic and hemorrhagic stroke. The interaction was tested by modeling the interaction between SBP and sex, age and BMI for stroke risk. For a sensitivity analysis, we additionally examined the shape of BP-stroke risk relation after winsorization of SBP and DBP at the 1st/99th. All statistical analyses were carried out using R software version 3.6.1 (http://www.R-project.org). All tests were 2 sided, and P < 0.05 was considered statistically signi cant.

Result
Baseline characteristics of 5,700 study participants are presented in Table 1. During the follow-up, 178 participants developed stroke events, including 65 ischemic strokes, 51 hemorrhagic strokes, and 62 unspeci ed stroke events.   We found no signi cant age, gender and BMI interaction for the association of SBP with overall stroke (P interaction = 0.48, 0.902, and 0.764 for age, gender and BMI with overall stroke) (Fig. 2).   (Table 3).

Association of BP level with ischemic stroke
Multivariable adjusted restricted cubic spline analyses (model 3) showed a linear association of SBP with ischemic stroke (P for linearity < 0.001); no signi cant linear and nonlinear relationships of DBP with ischemic stroke were observed (Fig. 1). Multivariable adjusted restricted cubic spline analyses (model 1 and model 2) showed linear association of SBP and DBP with ischemic stroke, respectively (Figure s1 and Figure s2 in appendix).
We found no signi cant age, gender and BMI interaction for the association of SBP with ischemic stroke (P interaction = 0.417, 0.664, and 0.626 for age, gender and BMI with ischemic stroke) (Fig. 2).   (Table 4).

Association of BP level with hemorrhagic stroke
Multivariable adjusted restricted cubic spline analyses (model 3) showed a linear association of SBP with hemorrhagic stroke (P for linearity 0.022); no signi cant linear and nonlinear relationships of DBP with hemorrhagic stroke were observed (Fig. 1). Multivariable adjusted restricted cubic spline analyses (model 1 and model 2) showed linear association of SBP and DBP with hemorrhagic stroke, respectively ( Figure   s1 and Figure s2 in appendix).
We found no signi cant age, gender and BMI interaction for the association of SBP with hemorrhagic stroke (P interaction = 0.532, 0.519, and 0.605 for age, gender and BMI with hemorrhagic stroke) (Fig. 2).

Discussion
We used data from the CHNS study, which was a nationally representative prospective cohort in nine provinces around China, enabling us to provide a general method to estimate the association of BP and risk of stroke for the Chinese population. In this study, increasing SBP was an independent risk factor and positively associated with overall stroke, ischemic stroke, and hemorrhagic stroke. The risks of ischemic and hemorrhagic stroke were higher when SBP was more than 140 mmHg and 160 mmHg( Table 2,  Table 3 and Table 4). Multivariable adjusted restricted cubic spline analyses showed linear association of SBP with overall, ischemic and hemorrhagic stroke (P < 0.001, P < 0.001 and P = 0.022).
A previous study proved that rising SBP and DBP levels were associated with the development of CVD [13]. Higher SBP level was positively associated with increased risk of CVD with adjustment or strati cation for DBP [14]. By contrast, another study demonstrated that DBP was not consistently associated with the risk of CVD, after adjusting or strati cation for SBP [15][16]. Similar to previous results, our ndings showed that SBP and DBP were positively associated with stroke in the multivariate model(model 1 and model 2). However, in a mutually adjusted model(model 3) including both SBP and DBP, the association with the risk of overall stroke was positive for SBP, but not signi cant for DBP( Table 2, Table 3 and Table 4).
In our study, the relation of SBP to the risk of overall stroke is continuous and graded and no critical value has been observed( Table 2, Fig. 1). Consistently, in another retrospective study with 1.25 million patients [17], a dose-response association for SBP with overall stroke had also been observed. Further, in a low-income population in China, signi cant linear associations of SBP to the risk of overall stroke had also been reported in a prospective cohort study [18]. In addition, a meta-analysis involved nearly 1 million participants and 56,000 vascular deaths from 61 prospective studies has also con rmed the continued and graded impact of BP on CVD risk [7]. This prospective research observed that BP was associated with vascular mortality, without any clear indication of a threshold. In the entire BP range, every 20 mmHg increase in SBP (or 10 mm Hg increase in DBP) will double the risk of stroke or coronary heart disease for people between 40 and 69 years of age. Furthermore, ndings from the Framingham study further suggested that there is a continuous, graded association of SBP to the rate of development of CVD at all ages, without any indication of a threshold was observed in any group from age 35 to 84 years [19]. Other studies have also reported similar linear associations of BP to coronary heart disease (CHD) and allcause mortality [20] [21].
In this study, compared to the reference group (SBP < 130 mmHg ), the ischemic and hemorrhagic stroke risks were signi cantly increased among those with SBP values ≥ 140 mmHg and ≥ 160 mmHg, after adjusting other conventional CVD risk factors(model 2 in Table 3 and Table 4). The ideal BP cut-off values for decreasing stroke risk were SBP values ≤ 140 mmHg and ≤ 160 mmHg for ischemic stroke and hemorrhagic stroke. These results were similar to a prospective cohort study [18], showing that stroke risk signi cantly increased among individuals with SBP levels ≥ 140 mmHg (for ischemic stroke) and ≥ 160 mmHg (for hemorrhagic stroke), compared with those with SBP levels < 130 mmHg. Further, in another retrospective study with 1.25 million patients [17], for SBP levels ≥ 160 mmHg, SBP was more strongly associated with hemorrhagic stroke than ischaemic stroke. The stroke subtypes for the association of BP and stroke risk needs to be demonstrated in further studies.
We found no signi cant age, sex and BMI interaction for the associations of SBP with ischemic stroke, hemorrhagic stroke, and overall stroke (Fig. 2). In sensitivity analysis, the shape of BP-stroke risk relation after winsorization of SBP and DBP at the 1st/99th did not change substantially (Figure s3 in appendix).
Our study has several limitations. First, we only screen residents older than 40 and younger than 90 years old from nine provinces of China. Therefore, our current results cannot be generalized to all populations in China. Second, the long-term, > 6 years, incidence of stroke for BP is unknown. In future studies, we project to prospective follow-up participants, gather data about mortality and incidence of stroke in participants. Finally, we did not collect information on usage of BP-lowering medication or treatment at baseline for participants from CHNS data.

Conclusion
In conclusion, our study found the relationship between BP and stroke subtypes in the Chinese general population. SBP is strongly and directly related to the risk of overall stroke, without any evidence of a threshold. Besides, the risk of ischemic, and hemorrhagic stroke might be higher when SBP were more than 140 mmHg and 160 mmHg. These results provide more evidence for guidelines or health policies of primary prevention of stroke subtypes with the management of BP level. Individual-level and populationlevel interventions to control BP level are needed to reduce the burden of stroke in China. instruments and the process for obtaining informed consent for the survey. All participants and/or their parents/guardians provided written informed consents for their participation in the survey.

Consent for publication
Not applicable.
Availability of data and materials The datasets generated and/or analysed during the current study are available in the CHNS repository, http://www.cpc.unc.edu/projects/china/data/datasets.

Competing interests
The authors declare that there is no con ict of interest.

Funding
Not applicable.
Authors' contributions QY and QDJ conceptualized and designed the study, carried out the initial analyses, drafted the initial manuscript, and reviewed and revised the manuscript; YQC, QYL and XRL critically reviewed and revised the manuscript; and all authors approved the nal manuscript for submission.

Figure 1
Association of BP with overall, ischemic and hemorrhagic stroke (model 3)

Figure 2
Association of BP with overall, ischemic and hemorrhagic stroke by age, gender and BMI (model 3)

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.