Short-Term and Long-Term Blood Pressure Changes and the Risk of All-Cause and Cardiovascular Mortality

Background Few studies compared the effects of BP changes in short- and long-terms on all-cause mortality and CVD mortality. Methods We performed a 12.5-year follow-up study to examine the association between short- (2008 to 2010) and long-term [baseline (2004-2006) to 2010] BP changes and the risk of mortality (2010 to 2017) in the Fuxin prospective cohort study. The Cox proportional hazards model was used for this study, and the average BP was stratified according to the Seven Joint National Committee (JNC7). Results We identified 1496 (805 CVD deaths) and 2138 deaths (1222 CVD deaths) in short- and long-term study. Compared with BP maintainer, in short-term BP changes, for participants from normotension or prehypertension to hypertension, the hazards ratios (HRs) and 95% confidence intervals (CIs) of all-cause mortality were 1.948 (1.118-3.392) and 1.439 (1.218-1.700), respectively, while for participants from hypertension to prehypertension, the HRs (95% CIs) were 0.766 (0.638-0.899) for all-cause mortality and 0.729 (0.585-0.908) for CVD mortality, respectively. In long-term BP changes, for participants from normotension or prehypertension to hypertension, the HRs (95% CIs) of all-cause mortality were 1.738 (1.099-2.749) and 1.203 (1.023-1.414), and they were 2.351 (1.049-5.269) and 1.323 (1.047-1.672) for CVD mortality, respectively. In addition, the effects of short-term BP changes on all-cause and CVD mortality, measured as regression coefficients (β), were significantly greater than those in long-term change (all P<0.05). Conclusions Our study emphasizes that short-term changes in BP have a greater impact on all-cause and CVD mortality than long-term changes and assess the cut-off value of the changes in blood pressure elevation.


Introduction
In 2017, the global adult mortality rate has declined slowly; not only that, but in some cases, the mortality rate is still rising [1]. Noncommunicable diseases accounted for 73% of the total number of deaths worldwide, more than half of which were attributed to only four risk factors, and hypertension was one of them [2]. The relationship between blood pressure (BP) level and mortality (all-cause or cardiovascular disease (CVD)) has been investigated in numerous studies [3][4][5][6][7]. It is well known that BP is an ever-changing variable in individuals during follow-up [8]. Therefore, office BP at a single point in time did not accurately predict all-cause mortality and researchers have begun to pay attention to the relationship between BP changes and death risk in recent years. For example, Fan JH et al. [9] found that, relative to stable BP of normotension, having a rise in BP from normotension or prehypertension to hypertension both conferred an increased risk of total and CVD and stroke mortality. Susanne M et al. [10] identified 4 BP trajectories and found that ten-year BP trajectories were the strongest predictors, among different BP measures, of CVD and all-cause mortality. In Kim MK's study [11], they found that the risk of cardiovascular outcomes was increased with greater variability in systolic blood pressure  (SBP) and greater BP variability leads to greater cardiac and vascular damage [12]. However, these studies focus on the impact of BP on outcomes at the same period, without considering both short-and long-term BP changes in one study. Short-and long-term BP changes may have different effects on mortality and it is unclear whether short-and longterm changes in BP categories are differentially associated with mortality risk. In our study, we focused more on the changes in BP, a particularly compelling and underreported putative all-cause mortality risk factor, and to evaluate the relationship between changes in BP categories and all-cause and CVD mortality in a representative natural population.

Study Population and Study
Design. This is a largescale epidemiological follow-up study. From 2004 to 2006, a multistage, random cluster sampling design was performed to select a representative sample of the rural population aged 35 years and older from Fuxin County of Liaoning Province. The detailed methodology was described elsewhere [13]. 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), investigators were invited to participate in the follow-up study. Of the 45,925 participants at baseline, 3,883 subjects missed contact information or refused to attend the follow-up, and 42,042 participants were eligible to attend the follow-up. Of these, 846 participants who were missing SBP, diastolic blood pressure (DBP), or other key variables (demographics, lifestyle, CVD disease history, and history of disease associated with stroke) at baseline were excluded. For short-term changes analyses, subjects with missing SBP, DBP, and other key variables and missing body mass index (BMI), current smoking, and current drinking at the followup 1 (n=10,219) and missing SBP and DBP at the follow-up 2 (n=6776) and who died before the follow-up 2 (n=197) were excluded, leaving 24,004 participants for analysis. For long-term changes, subjects with missing SBP, DBP, and other key variables at the follow-up 2 (n=10,402) and who died before the follow-up 2 (n=288) were excluded leaving 30,506 participants for analyses ( Figure 1) Figure 2: Study designs used for analysis of the associations between short-and long-term changes.
followed were in accordance with the ethical standards of the responsible committee on human experimentation of China Medical University, and written informed consent of all subjects or their agents was obtained.
We aim to compare the association between short-and long-term BP changes and all-cause and CVD mortality ( Figure 2). Long-term BP changes were between baseline and follow-up 2, and short-term BP changes were between followup 1 and follow-up 2. Data on events were all collected from 2010 to follow-up 2.

Baseline
Measurement. Data on demographic variables (age, sex, and race), current drinking [14], current smoking, physical activity, history of disease (stroke, coronary heart disease (CHD), family history of hypertension, diabetes, and hyperlipemia), and information on antihypertensive medications were obtained by interview with a standard epidemiological questionnaire.

Follow-Up.
All subjects were invited to attend the followup. A total of 42,242 patients finished at least one time follow-up (follow-up rate 91.5%). At each visit, we collected the information on clinical end points and concurrent medication use. During each visit, three BP measurements were taken according to a standard protocol identical to that of the baseline examination. The mean value of three BP measurements was used for each participant. We then evaluated the risk of study outcomes according to BP.

Study Outcomes.
Our results included all-cause and CVD mortality. Deaths were confirmed through hospital records and direct contact with their families. We confirmed that death from CVD on the basis of autopsy reports, death certificates, medical record abstract, or information obtained from family members [16]. All materials were independently reviewed by the end-point assessment committee which included the certified neurologists, cardiologists, and others.

Statistical
Analysis. Continuous variables were reported as means and standard deviations (SD), and categorical variables were expressed as frequency and percentage. The rates of events were presented as the number of events per 1000 person-years. We used multivariable Cox proportional hazards models to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for the associations between BP categories and mortality. We calculated the risk of events for participants with altered BP levels, with reference to those with unchanged BP categories. Next, we calculated the events risk of participants with changes in BP levels and made normotension BP as a reference.
We adjusted for sex, age, race, BMI, SBP, DBP, current smoking, current drinking, education level, physical activity, antihypertensive treatment, family history of hypertension, and history of diabetes, hyperlipidemia, and CVD. Beyond that, we compared the predictive power of short-term changes and long-term changes using Fisher Z test [17]. A 2-sided P value <0.05 was deemed significant. Moreover, receiver operating characteristic (ROC) curves were constructed, and the area under the curve (AUC) was calculated to assess the cut-off value of the SBP changes in BP elevation. All analyses were performed with SPSS statistical software version 20.0 (SPSS Inc., Chicago, Illinois, USA). A P value less than 0.05 was accepted as indicating statistical significance.

Results
There were 24,004 participants in short-term analysis and 30,506 participants in long-term analysis, of which 49.4% and 51.1% were women, and the mean age was 51.9 (SD, 10.8) years and 50.2 (SD, 11.0) years, respectively.  Figure 3 shows the number of cases of all-cause mortality ( Figure 3(a)) and CVD mortality (Figure 3(b)) per 1000 person-years by BP categories change. In the short-term BP changes study, 1496 deaths (including 805 from CVD deaths) were identified, and the overall incidence of all-cause mortality was 6.97 per 1000 person-years (CVD mortality was 3.75 per 1000 person-years). For the long-term BP changes analysis, there were 2138 all-cause mortality (1222 CVD mortality), and the total incidences of all-cause mortality and CVD mortality were 7.83/1000 person-years and 4.47/1000 person-years, respectively. Table 2 shows HRs (95%CI) for the associations between BP changes and risk of all-cause and CVD mortality. In BioMed Research International 5      the multivariate adjusted Cox model for all-cause mortality, compared with BP maintainers, in short-term BP analysis, we found a significant decreasing risk of BP from hypertension to prehypertension, and the HRs (95%CI) were 0.766 (0.638-0.899) for all-cause mortality and 0.729 (0.585-0.908) for CVD mortality, respectively. And the results of women were the same; the HRs (95%CIs) were 0.606 (0.448-0.822) for allcause mortality and 0.610 (0.418-0.889) for CVD mortality, respectively. In contrast, there were significant increasing risks of BP categories changes, and participants with BP from normotension or prehypertension to hypertension had HRs (95% CI) of 1 We also compared the effects of short-and long-term BP changes, measured as regression coefficients ( ), and they were significantly greater in short-term changes than in long-term for all-cause mortality (total: =0.667 VS =0.553, In Table 3, we compared all the other 8 groups with the group 1 as reference. In short-term changes, from normotension to hypertension increased significantly for allcause mortality, the HRs (95% CI) were 1.846 (1.092-3.181 For participants with elevated SBP, the cut-off value of SBP changes was evaluated by using the ROC curve to predict mortality (Figure 4). In short-term analysis (Figure 4(a)), the optimal cut-off value of SBP changes for the diagnosis of allcause mortality was 11.5 mmHg in total, and the AUC (95% CIs) was 0.538 (0.517-0.559). And the cut-off value was 13.5 mmHg for CVD mortality; the AUC (95% CIs) was 0.543 (0.513-0.573). Among men, the cut-off values of all-cause and CVD mortality were 18.5 mmHg and 13.5 mmHg, and the AUC (95% CIs) were 0.543 (0.516-0.570) and 0.548 (0.510-0.586). Among women, for all-cause mortality, the cut-off value was 11.5 mmHg, and the AUC (95% CIs) was 0.548 (0.512-0.583). For CVD mortality, the cut-off value was 12.5 mmHg, and the AUC (95% CIs) was 0.549 (0.500-0.598). In long-term analysis, the cut-off values of all-cause mortality were 18.5 mmHg (AUC: 0.529; 95% CIs: 0.509-0.548) and they were 19.5 mmHg (AUC:0.538; 95% CIs:0.511-0.564) of CVD mortality for total. Among men, the cut-off values of allcause and CVD mortality were 19.5 mmHg (AUC:0.538; 95% CIs:0.514-0.563) and 21.5 mmHg (AUC:0.562; 95% CIs: 0.530-0.595), respectively. The results of nondiabetics are shown in Supplementary Appendix 3.

Discussion
The main findings of the present study were the positive association between short-and long-term BP changes and the risk of all-cause and CVD mortality in rural areas of China. Overall, our data showed that, compared with people who maintain the BP status, participants with elevated BP had a high risk of all-cause and CVD mortality, and participants with reduced BP had a lower risk of all-cause and CVD mortality, both in short-and long-term changes analysis. In addition, the difference between short-term and long-term changes is statistically significant.
Our study confirmed the findings for positive associations between BP changes and all-cause mortality, which was comparable with previous studies [18][19][20][21]. Two studies were performed on hypertensive patients and untreated hypertensive patients at the IPC Center in Paris, and changes in individual long-term BP are independent predictors of allcause mortality in hypertensive patients [18]. Data from the Minnesota Business and Professional Men Study (n=261) and the Zutphen Study were shown, and the 10-year BP trajectory was the strongest predictor of cardiovascular mortality and all-cause mortality in Minnesota [19]. Cardiovascular health studies concluded that long-term visit-to-visit SBP variability was independently associated with a higher risk of subsequent mortality and a meta-analysis of 13 cohort studies in Japan also presented that adjusted mortality increased SBP changes for All-cause mortality in Total SBP changes for All-cause mortality in Men SBP changes for CVD mortality in Total SBP changes for CVD mortality in Women SBP changes for CVD mortality in Men   with increasing BP [20,21]. These studies found that BP changes are a powerful predictor of cardiovascular events independently of mean SBP or DBP, which is more common in previous studies. And we also got the same conclusion when studying CVD and all-cause mortality. In addition, we explored the relationship between BP changes in the longitudinal pattern over time and the risk of subsequent mortality. What is more, we used BP short-and long-term changes to explore the relationship between BP and mortality, and this research was still very scarce in China. We believe that short-and long-term changes in BP are an independent risk factor of mortality. And the impact of short-and long-term changes on outcomes is different. BP changes are indeed the result of a complex interaction between external environmental stimuli and the response of several cardiovascular control mechanisms [22]. There is evidence that short-term BP changes predict terminal organ damage and cardiovascular events [23][24][25]. However, the individual biologic mechanisms by which long-term BP changes may affect risk of mortality CVD or all-cause mortality are yet unclear. This impact might be due to changes in BP that cause significant changes in vasculature exposure to pressure load over a long period of time [26]. These would affect the potential health of vascular tissue, thereby affecting the development or severity of CVD [27,28]. Therefore, one hypothesis is that short-and long-term changes are different for vascular pressure states. Further studies are warranted to test for this hypothesis. We detected that the same changes in BP occur, and the shorter change time seems to affect mortality more. The content of the appeal reminds us that if only one baseline BP measurement is used, the impact of hypertension on the outcome of the event will be overestimated or underestimated! Therefore, pay special attention to sudden changes in BP. Only in this way can we better help prevent cardiovascular and all-cause mortality.
Interestingly, a declining BP in hypertensive patients increases the risk of all-cause mortality, compared with maintaining normotension, especially for CVD mortality. Some studies showed that a low DBP was associated with an increased all-cause mortality risk [29,30]. In post hoc analyses of the Systolic Hypertension in the Elderly Program (SHEP), after fully adjusting the functional status and other confounding factors, the drop in BP is still accompanied by an increase in mortality. It is also possible that lower BP in the elderly may increase the risk of adverse outcomes [31]. In assessing the prevalence of BP decline in the elderly and its relationship to subsequent outcomes, Satish S et al. [32] found that a drop in BP may be a predictor of higher mortality risk in the elderly. This result appeared more in the elderly as shown in previous studies, but the reason is unclear.

Strengths and Limitations.
Our study had several important strengths, which include the relatively large sample size and large number of adverse events accrued, thereby increasing the statistical power of our analyses. Moreover, in this study, information about BP was derived from the mean BP of follow-up, which helped determine the relationship between BP and adverse outcomes. Finally, in addition to looking at the effect of changes in BP Categories on death, we used the ROC curve to find the cut-off value of SBP in patients with elevated BP. Our study also had several limitations. First, the BP in this study is the average of three measurements in a day, so we might not account more for the effects of individual BP fluctuations. Secondly, our research sample only included participants in rural areas of China, and we could expect different results from more people of different ethnicity. Thirdly, we did not have enough laboratory measurements, such as cholesterol, blood glucose, and inflammatory biomarkers to control these covariates. This is why we have the low frequency off diabetes in baseline characteristics. Finally, it is also worth noting that the classification of BP changes may mask some of the individual variabilities of BP in terms of time variation and may lead to attenuation.

Conclusions
In our study, using the Cox proportional hazard models with short-term BP changes and long-term BP changes entered in the same model, BP changes provided more information on risk of all-cause and cardiovascular mortality than BP at a single point in time. Our research suggests that short-term BP changes have a greater effect on mortality. And individuals who are able to maintain their BP to normal BP levels have the lowest risk for CVD and all-cause mortality. The importance of hypertension management should be widely accepted in public health practice. Prevention efforts should continue to emphasize the importance of lowering BP and maintaining normotension to reduce the mortality.

Data Availability
The data used to support the findings of this study are available from the corresponding author upon request.

Disclosure
There are no relationships with industry.