Short‐term blood pressure changes have a more strong impact on stroke and its subtypes than long‐term blood pressure changes

Abstract Background Elevated blood pressure (BP) is closely related to stroke and its subtypes. However, different time periods changes in BP may result in differential risk of stroke. Hypothesis Short‐term blood pressure changes have a more strong impact on stroke and its subtypes than long‐term blood pressure changes. Methods We designed the study on the effects of short‐ (2008‐2010) and long‐term (2004‐2010) BP changes on stroke events (2011‐2017), including 22 842 and 28 456 subjects, respectively. The difference in β coefficients between short‐ and long‐term BP changes on the effects of stroke were examined using the Fisher Z test. Results During a median 12.5‐year follow‐up period, 1014 and 1505 strokes occurred in short‐ and long‐term groups. In short‐term group, going from prehypertension to hypertension, the risk of stroke events increased (stroke: hazard ratio [HR] = 1.537 [1.248‐1.894], ischemic stroke: 1.456 [1.134‐1.870] and hemorrhagic stroke: 1.630 [1.099‐2.415]); going from hypertension to prehypertension, the risk of stroke events decreased (stroke:0.757 [0.619‐0.927] and hemorrhagic stroke:0.569 [0.388‐0.835]). Similarly, in long‐term group, going from prehypertension to hypertension, individuals had an increased risk of stroke (1.291, 1.062‐1.569) and hemorrhagic stroke (1.818, 1.261‐2.623); going from hypertension to prehypertension, participants had a decreased risk of stroke (0.825, 0.707‐0.963) and hemorrhagic stroke (0.777, 0.575‐0.949). Furthermore, the effects of BP changes during short‐term period on stroke events were greater than that in long‐term period. Conclusions Short‐ and long‐terms BP changes were both associated with the risk of stroke events. Furthermore, short‐term BP changes had a stronger impact than did long‐term changes on risk of stroke events.


| INTRODUCTION
Stroke is the leading cause of death in China, 1,2 and high blood pressure (BP) is a major risk factor for stroke. 3 Worldwide, approximately 54% of strokes are attributable to hypertension. 4 Hence, lowering BP can effectively reduce the incidence of stroke. In the past decade or so, the prevalence of hypertension in rural areas of northeast China has changed considerably, 5,6 and the impact of this rising trend on the incidence, mortality of stroke is worrisome.
In addition, previous traditional studies have been linked to the association of BP values at one point in time with cardiovascular disease (CVD) risk. However, intraindividual BP fluctuates dynamically over time, and measurements at a single point in time may not effectively predict disease risk, to achieve early intervention. 7 Gosmanova et al determined that traditional analysis would underestimate the true relationship between elevated BP levels and disease, as a result of a "regression dilution" effect. 7 Vascular damage leading to stroke is a complex dynamic process. Dynamic BP changes include more critical information for prediction of stroke risk, having higher accuracy and sensitivity. 8 Several lines of evidence that visit-to-visit variability (VVV) of BP or trajectory changes in BP are associated with a higher incidence of CVD 9-12 and a higher risk of mortality. [13][14][15] However, few studies have concentrated on the effects of BP changes on stroke and its subtypes, and most of those studies were conducted either in Western countries 7,10 or focused on the adverse outcomes affected by BP changes during the same time period 16,17 or same multiple time periods. 9 There is rare evidence of the effects of short-and long-term BP changes on stroke and its subtypes in the general population, while the effects of long-and short-term BP changes may be different.
We hypothesized that BP changes will affect to stroke and its subtypes according to three aspects: first, baseline BP levels are positively correlated with the risk of stroke for the same degree of BP changes; second, the direction of BP change (increase or decrease) is an important aspect of responding to BP changes; third, the rate of BP changes (high-increasing or low-increasing, high-decreasing or low-decreasing) have different effects on stroke in the same BP level changes. Therefore, we aimed to analyze and compare the effects of same level BP changes over a short-and long-term period on the risk of stroke and its subtypes.

| Study population
We carried out a large-scale epidemiological follow-up study. 18,19 Briefly, from 2004 to 2006, a multistage, random cluster sampling process was used to select a representative sample aged ≥35 years in rural areas of Fuxin County in Liaoning Province, China. All study participants were invited to return for follow-up: from January to July 2008 (follow-up 1); from July to December 2010 (follow-up 2); and from March to December 2017 (follow-up 3). The study population inclusion and exclusion process is illustrated in Figure S1. Of the 45 925 participants at baseline, 3883 subjects had missing contact information or refused to attend the follow-up, and 42 042 (91.5%) participants were eligible to attend the follow-up at least one time.
The study protocol was approved by the China Medical University Research Ethics Committee, and written informed consent was obtained from all subjects or their caregivers.

| Study design
We have completed a baseline survey and three follow-up surveys ( Figure 1A). To investigate the effects of BP changes over short-and long-term time intervals on stroke events, we collected new cases of stroke during the same time period between the short-and long-term groups. Therefore, the study designs used for analysis of the associations between long-(from January 2004 to December 2010) and short-term (from January 2008 to December 2010) BP changes and stroke events is shown in Figure 1B,C.

| Data collection and measurements
Using a standardized questionnaire to collect demographic variables (gender, age, and ethnicity), lifestyle factors (current smoking, current drinking status, and physical activity), history of disease (stroke, coronary heart disease [CHD], family history of hypertension, diabetes, or hyperlipidemia), and information on antihypertensive medications. 18,19 When measuring weight and height, subjects did not wear coats and shoes. Body mass index (BMI) was calculated as the weight divided by the height squared (kg/m 2 ). BP measurements were performed using a standard electronic automated sphygmomanometer (HEM-741C; Omron, Tokyo, Japan). A trained and certified observer used an American Heart Association protocol to perform BP measurements. 20 Participants were advised to avoid alcohol consumption, cigarette smoking, drinking coffee/tea, and exercise for at least 30 minutes, and to rest at least 5 minutes before BP measurement.
The individual's BP value used for analyses was the average of three BP measurements. According to the JNC7 criteria, 21

| Statistical methods
Data were presented as the means ± SD, and number (percentages).  In terms of the overall trend, subjects who were able to maintain normotension or decrease their BP to normotension had a lower incidence rate in shortand long-term groups. Meanwhile, we found that normotensive subjects in short-term group presented higher incidence rate of the three outcomes than long-term group, no matter if they remained normotensive or became prehypertensive or hypertensive. While the contrary happen when considering hypertensive subjects in long-term vs shortterm group. Furthermore, after short-and long-term BP changes, the incidence rate of all the other eight categories was higher than that for maintaining the normotension. the whole population in short-and long-term groups ( Table S1). In addition, β coefficients from Cox proportional hazards models were used to compare the effects of short-and long-term BP changes on the risk of stroke and hemorrhagic stroke. For individuals whose BP changed from prehypertension to hypertension, short-term changes were more strongly associated with increased risk of stroke than were long-term changes on the basis of β coefficients (β = 0.430 vs 0.255, P < .001). Similarly, for individuals whose BP changed from hypertension to prehypertension, the β coefficient of short-term BP changes was also higher compared to long-term (β = −0.278 vs −0.193, P < .001). The opposite occurred in hemorrhagic stroke: for individuals whose BP changed from prehypertension to hypertension, long-term changes were more strongly related to increased risk of hemorrhagic stroke compared to short-term changes on the basis of β coefficients (β = 0.598 vs 0.488, P < .001). Conversely, for individuals whose BP changed from hypertension to prehypertension, the β coefficient of short-term BP changes was higher compared to long-term (β = −0.564 vs −0.253, P < .001). Figure 3 shows the cumulative incidence risk of long-and short-term BP changes to stroke. Regardless of short-or long-term BP changes, individuals who were able to maintain or to decrease their BP values had a lower cumulative incidence of stroke. In contrast, those who had higher initial BP values or an increase in their BP values had a higher of cumulative incidence of stroke. The cumulative incidence of stroke for all other eight categories was higher than that for maintaining the normotension, and individuals who maintained the hypertension and normotension had the highest and lowest cumulative incidence, respectively.
F I G U R E 2 The incidence of stroke and its subtypes with short-term and long-term blood pressure changes. Stroke, A; ischemic stroke, B; hemorrhagic stroke, C; Nor, normotension; Pre, prehypertension; Hyp, hypertension. Error bars represent 95% CI

| DISCUSSION
We studied and compared the effects of BP changes over a shortand long-term period on stroke and its subtypes in a prospective cohort study in Chinese rural adults, and found that the majority of  β Coefficients different from 0 for short-and long-term blood pressure changes groups in stroke: P < 0.05. c β Coefficients different from 0 for short-and long-term blood pressure changes groups in hemorrhagic stroke: P < 0.05. Bold values in Table 2  In addition, with the hypertension as reference, individuals who changed from hypertension to prehypertension decreased the risk of stroke events in short-and long-term groups. However, there was no statistical significance in the risk of stroke events in BP classifications of individuals who changed from hypertension to normotension compared with maintained hypertension. This is maybe a power issue (fewer people make the changing from hypertension to normotension), or it could also imply a J-curve effect in lowering BP. 29 Although some of the results in our subgroup analysis have no statistical significance, they cannot be ignored. We still cannot deny that antihypertensive treatment can reduce the risk of the stroke events.
Our results may be by chance because of the small number of cases and insufficient statistical power. Cumulative BP load is related to the risk of stroke in a later time period. In short-and long-term BP change analyses, we found that initial BP level at a higher BP classification or more rapid increase in BP was significantly correlated with the cumulative incidence of stroke.
Our results support previous findings of observational studies, which suggested that increase in BP is detrimental to health. A study of seven diverse US cohort studies estimated how BP changes affect the lifetime risk for CVD and found that increases in BP showed a positive relationship with remaining lifetime risk for CVD. 10  More study population would be expected to predict the associations between BP changes and the risk of stroke. Fifth, information on the antihypertensive treatment of hypertensions in our study is limited.
More information about hypertension population with or without antihypertensive treatment is needed to as the potential confounders to confirm our findings.
In conclusion, our study found that short-and long-term changes in BP differentially affect the subsequent risk of stroke events independent of baseline BP levels and that elevated BP would increase the future risk of stroke and its subtypes. Furthermore, decreased BP can lower the risk of incidence but does not reverse the persistent negative effects of exposure to elevated BP. Even so, great attention should be paid to increasing the rate of hypertension control in rural areas of China.

ACKNOWLEDGMENTS
This work was supported by funds from National Nature Science Foundation of China (No. 81773510) and National Key R&D Program of China (Grant Nos. 2017YFC1307600, 2018YFC1311600). All of the investigators and staff members were gratefully acknowledged. Thanks for all the enthusiastic participants. Especially, we thank Professor Yanan Ma for giving us guidance during the process of data collation and analysis.