Time‐weighted blood pressure with cardiovascular risk among patients with or without diabetes

Abstract Background Usual measures of blood pressure (BP) do not account for both the magnitude and duration of exposure to elevated BP over time. We aimed to demonstrate the effect of a novel time‐weighted BP on cardiovascular outcomes using a post hoc analysis of two published randomized trials. Hypothesis Time‐weighted blood pressure is associated with cardiovascular risk among patients with or without diabetes. Methods The limited‐access ACCORD and SPRINT data sets were used for the current study. Time‐weighted BP is obtained by dividing cumulative BP by the total follow‐up time. Time‐weighted BP burden above a threshold is also determined after deriving the time‐weighted BP by re‐zeroing the interpolated pressure values at two different hypertension thresholds (>140/90 and >130/80 mmHg). Results Eighteen thousand five hundred forty‐one patients from the two clinical trials were enrolled in this study. A J‐curve relation was observed between time‐weighted BP and major cardiovascular events (MACE). The systolic blood pressure (SBP) burden independently predicted MACE across the two trials at different thresholds (ACCORD: SBP > 130 mmHg, HR = 1.05 [1.03−1.06]; SBP > 140 mmHg, HR = 1.06 [1.04−1.08]; SPRINT: SBP > 130 mmHg, HR = 1.04 [1.03−1.05]; SBP > 140 mmHg, HR = 1.05 [1.04−1.07]). Consistent results were found for diastolic blood pressure (DBP) burden (ACCORD: DBP > 80 mmHg, HR = 1.10 [1.06−1.15]; DBP > 90 mmHg, HR = 1.20 [1.11−1.30]. SPRINT: DBP > 80 mmHg, HR = 1.06 [1.02−1.09]; DBP > 90 mmHg, HR = 1.12 [1.06−1.18]). Significant associations were also observed for stroke, myocardial infarction, cardiovascular death, and all‐cause mortality. Conclusion Both time‐weighted SBP and DBP independently influenced the risk of adverse cardiovascular events among patients with and without diabetes, regardless of the definition of hypertension (130/80 or <140/90 mmHg).


Funding information
National Natural Science Foundation of China, Grant/Award Number: 82370387 Conclusion: Both time-weighted SBP and DBP independently influenced the risk of adverse cardiovascular events among patients with and without diabetes, regardless of the definition of hypertension (130/80 or <140/90 mmHg).

K E Y W O R D S
blood pressure, cardiovascular events, diabetes, hypertension

| INTRODUCTION
2][3][4] Because the management of hypertension has traditionally centered around BP measurements taken at a single time point, it fails to recognize BP as a continuous measure that fluctuates over time.][10] The cumulative BP integrated both the magnitude and duration in units of mm Hg × time.However, this measure cannot differentiate the same cumulative BP in the form of a very high BP level with a short follow-up and a high BP level with a long follow-up.
In acute HF patients, 24-hour time-weighted lactate had a greater predictive value than the other static and dynamic indices of lactate homeostasis. 11Similarly, time-weighted BP is obtained by dividing cumulative BP by the total follow-up time, which takes into account the amount of time spent at every single BP in relation to the total period of time observed. 9,12,13As the predictive value of time-weighted BP remains elusive, we aim to explore the effect of time-weighted systolic blood pressure (SBP)   and diastolic blood pressure (DBP) on the risk of major cardiovascular events (MACE) in patients with or without diabetes on the basis of a post hoc analysis of two published randomized trials, ACCORD 14,15 (Action to Control Cardiovascular Risk in Diabetes) and the SPRINT 16,17 (Systolic Blood Pressure Intervention Trial).
Our objective is to gain a deeper understanding of the impact of time-weighted SBP and DBP on the risk of MACE and to compare these effects in patients with and without diabetes by conducting this study.This research will advance the knowledge of the role dynamic BP measurements play in the development and prognosis of cardiovascular diseases, which will help to improve precise guidance for personalized hypertension management.The findings of this study may contribute to enhancing BP management strategies, which would ultimately lower the prevalence of cardiovascular diseases and their relevant adverse effects.

| Trial design and oversight
The limited-access ACCORD and SPRINT data sets obtained from the NIH upon approval were used for the current post hoc analysis.5][16][17] In general, the ACCORD trial was a double two-bytwo factorial, parallel treatment trial in 10 251 patients with type 2 diabetes.All participants in the ACCORD were randomly assigned to one of two glycemic treatment arms, with the standard arm targeting Additionally, both trials were approved by the institutional review board or ethics committee at each study site and all participants provided written informed consent.This post hoc analysis was waived for ethical approval by the ethical committee of the Liverpool School of Tropical Medicine (No:20-077).In our current analysis, we excluded 887 patients from the ACCORD trial and 184 patients from SPRINT trial because of missing data or lost follow-up.

| BP measurement
In both trials, BP was measured by the automated device (Model 907; Omron Healthcare) while the participant was seated.In the SPRINT trial, BP measurements were taken monthly in the first 3 months and every 3 months thereafter.In the ACCORD trial, patients allocated to the intensive BP treatment group were seen every a Missing value of time-weighted blood pressure has been deleted from current analysis.
b Time-weighted blood pressure is used to estimate mean blood pressure.c The population with a past medical history of stroke or heart failure has been excluded from the SPRINT trial.

| Outcomes
The primary outcomes for the ACCORD and SPRINT trials, adopted in our analysis, were the first occurrence of a MACE.
For the ACCORD trial, the definition of primary outcome was  and DBP burden (≥80/90 mmHg) as natural cubic splines to account for a continuous nonlinear functional dependence.We specified 0 mmHg as the reference value for SBP and DBP burden.Spline knots were placed at the 5th, 50th, and 95th centiles of the overall distribution of BP burden.
We also performed a series of sensitivity analyses including analyses were performed using STATA version 15.0 (Stata Corporation).

| RESULTS
A total of 18 541 participants from the two clinical trials were assessed in this cohort study.Among the participants, White individuals constituted the highest proportion (60.06%), followed by Black individuals (24.37%).Table 1 shows the baseline characteristics for the participants from the subgroups of BP burden (≥140/90 or ≥130/80 mmHg).In general, baseline characteristics in higher BP burden tend to be older age, also a higher percentage of black race, stroke, and HF history.
In general, a J-curve relation was seen between time-weighted SBP or DBP and the composite outcome.In both the ACCORD and SPRINT cohorts, the incidence rate began to increase when SBP exceeded 130 mmHg and DBP exceeded 80 mmHg.Interestingly, in the SPRINT cohort, there was a significantly faster increase in the risk of cardiovascular events following SBP exceeding 130 mmHg compared to the ACCORD cohort.This could be attributed to the fact that individuals with diabetes are inherently more susceptible to cardiovascular events (Figure 1).Our results from the Cox model further demonstrated that the SBP burden (≥140 mmHg) significantly increased the incidence of MACE, stroke, MI, HF, CVD death, and all-cause mortality in both ACCORD and SPRINT trials (Table 2).The DBP burden (≥90 mmHg) was also independently associated with the composite outcome, HF, CVD death, and all-cause mortality in the two trials.What's more, the association between a high DBP burden (≥90 mmHg) and stroke was found to be statistically significant only in the ACCORD trial.Similarly, the statistically significant association between a high DBP burden (≥90 mmHg) and MI was observed only in the SPRINT trial (Table 3).When we set the threshold for BP burden at 130/80 mmHg, the results are consistent with the aforementioned findings.The sensitivity analysis results support the stability of the research findings (Appendix Tables S1 and S2).There was no statistically significant correlation that appeared to be found between the incidence rate of MACE and the last BP measured before MACE or baseline BP, T A B L E 3 Associations between DBP (≥80/90 mmHg) burden and cardiovascular events in the ACCORD and SPRINT trials.with the exception of DBP (≥80 mmHg), and this was consistent across both cohorts (Appendix Table S3).

| DISCUSSION
This study was to investigate the prognosis value of time-weighted BP in a hypertensive population with or without diabetes.We found that the rate of MACE increased from the time-weighted SBP ≥ 130 mmHg subgroups and time-weighted DBP ≥ 80 mmHg subgroups.A J-curve relation was observed between time-weighted BP and MACE.These findings were similar between patients with or without diabetes, which could support tightened BP targets (<130/80 mmHg) for hypertensive patients.
Repeated assessments of BP, rather than a single baseline measure, would provide a more accurate representation of an individual's BP profile. 5,18,19Even in a well-controlled hypertensive population, a normal distribution of BPs around the mean is a common phenomenon, while the MACE risk is determined by both the magnitude and the cumulative duration of exposure to high BP. 20Ambulatory BP measurement predicts fatal and nonfatal MI and stroke better than standard office measurement does. 5Day-to-day (office visit-to-visit) variability in BP has been proposed as an independent predictor of cardiovascular risk. 6In a recent study, cumulative SBP load was recognized as a superior predictor of MACE compared with mean SBP, SBP time at target, and SBP SD among patients with type 2 diabetes. 18In our study, both timeweighted SBP and DBP were found to be efficient for cardiovascular risk prediction.This novel index describes not only the magnitude but also the duration and trend over the time of BP monitoring.
7 In a recent study using data from 1.3 million adults in a general outpatient population, both SBP and DBP independently influenced the risk of MACE, regardless of the definition of hypertension (≥140/90 or ≥130/80 mmHg).28 Our results also showed that time-weighted SBP and DBP each independently influenced cardiovascular outcomes, and therefore DBP ought not to be ignored in patients with or without type 2 diabetes.

| Limitations
The majority of the population in this study is from northern America.
Additional external data is required to validate the results of our study.Potential randomized clinical trials guided by time-weighted BP would be attested to improve unacceptable low BP control.

| Future directions
There are numerous pathways for researching and exploring further in this field, although the current study yielded compelling evidence concerning the predictive value of time-weighted BP for cardiovascular events.Initially, it would be vital to conduct longer-term followup studies with larger sample sizes to assess the sustained predictive capability of this measurement.Monitoring participants for a vast period would enable an ample assessment of the long-term consequences of time-weighted BP on cardiovascular outcomes.

HbA1c 7 .
0%−7.9% and the intensive targeting HbA1c < 6.0% (ACCORD glycemia trial).Half of the participants (5518) of the ACCORD participants were also allotted to a lipid intervention consisting of randomization to fenofibrate versus placebo (ACCORD lipid trial).The other half (4733) of participants were either intensive (SBP ≤ 120 mmHg) or standard (SBP ≤ 140 mmHg) BP treatment (ACCORD-BP trial).The mean duration of follow-up was 3.7 years in the ACCORD trial.The SPRINT study was a randomized, openlabel, multicenter trial.Nine thousand three hundred sixty-one highrisk patients were assigned to either intensive or standard BP treatment group similar to those used in the ACCORD-BP trial, with a median follow-up of 3.26 years.In contrast with the ACCORD trial, the participants in the SPRINT study excluded those with diabetes.

d
Figure S1).The 6-month visit time point was chosen because of the stability of BP values after this visit.
nonfatal MI, nonfatal stroke, or cardiovascular death.For the SPRINT trial, the MACE was defined as MI, acute coronary syndrome not resulting in MI, stroke, acute decompensated HF, or cardiovascular death.The secondary outcomes in our analysis included stroke, MI, HF, cardiovascular death, and all-cause mortality.F I G U R E 1 Adjusted relationships between time-weighted BP (categorical variables) and incidence rate of primary outcome in the SPRINT and ACCORD trials.The relationships were estimated by using Poisson regression after adjusting the variables of age, sex, race, treatment group, history of clinical CVD, history of dyslipidemia, history of hypertensive treatment, current smoking, current drinking, BMI, baseline SBP, eGFR, glucose, HDL-C, LDL-C.The Lowess method (the shadow represents the upper and lower bounds of 95% CI) was used to connect the incidence rate of the primary outcome in each time-weighted SBP or DBP subgroup.BMI, body mass index; BP, blood pressure; CVD, cardiovascular disease; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; SBP, systolic blood pressure.

F I G U R E 2
Spline analyses of SBP (≥130/140 mmHg) and DBP (≥80/90 mmHg) burden in the SPRINT and ACCORD trials.HRs for the combined primary outcome (shadow represent upper and lower bounds of 95% CI) is relative to 0 mmHg for SBP and DBP burden.Knots are placed at the 5th, 50th, and 95th centiles of the time-weighted SBP and DBP.Multivariable model was adjusted for the variables of age, sex, race, treatment group, history of clinical CVD, history of dyslipidemia, history of hypertensive treatment, current smoking, current drinking, BMI, baseline SBP, eGFR, glucose, HDL-C, LDL-C.BMI, body mass index; CVD, cardiovascular disease; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; SBP, systolic blood pressure.population.The landmark clinical trial ACCORD showed a similar cardiovascular risk between target SBP of <120 mmHg and target SBP of <140 mmHg in patients with diabetes.14Furthermore, the 2017 ADA Position Statement on Diabetes and Hypertension does not promote a uniform BP target (<130/80 mmHg) and instead risk stratifies to avoid overtreatment in type 2 diabetes patients.24Our results demonstrated a J-curve relationship between time-weighted BP and MACE with nadirs at SBP 130−135 mmHg or DBP 80−85 mmHg, which may support the recommendation of a target of BP < 130/80 mmHg.
Our study proposed a novel assessment of time-weighted SBP and DBP for predicting the risk of MACE.Both time-weighted SBP and DBP independently influenced the risk of MACE among patients with and without diabetes, despite the two distinct definitions of hypertension (defined as BP ≥ 140/90 or ≥130/80 mmHg).The findings of this study demonstrate the importance of adequate and sustained control of both SBP and DBP throughout life for prevention of cardiovascular events.

1
Baseline characteristics of study population in the ACCORD and SPRINT trial.a Categorical variables are reported as percentages of the characteristic.Continuous variables are reported as mean ± SD.
Associations between SBP (≥130/140 mmHg) burden and cardiovascular events in the ACCORD and SPRINT trials.
Note: Model 1: adjusted for age, sex, race, current smoking, current drinking, BMI; Model 2: adjusted for age, sex, race, treatment group, history of clinical CVD, history of dyslipidemia, history of hypertensive treatment, current smoking, current drinking, BMI, baseline SBP, eGFR, glucose, HDL-C, LDL-C.