Integrating longitudinal blood pressure history with current in-office measurements for cardiovascular risk stratification: a nationwide cohort study
K Ejiri, T Seki, M Okada, K Miyake, K Kubota, T Miyoshi, Y Kawazoe, S YuasaAbstract
Background
Current clinical guidelines recommend repeated in-office blood pressure (BP) measurements for accurate hypertension screening. However, the prognostic utility of integrating historical BP trajectories with current in-office BP readings for the risk stratification of future cardiovascular disease (CVD) remains unclear.
Objective
This study aimed to investigate whether incorporating longitudinal BP history with cross-sectional in-office BP measurements enhances risk prediction for incident CVD, utilizing a large-scale health checkup database in Japan.
Methods
We conducted a retrospective cohort study using the nationwide JMDC Claims Database (2005–2024). The cohort comprised 966,904 adults (≥18 years; median age, 45 [interquartile range (IQR), 37–51] years; 71.3% male) who underwent five consecutive health checkups and were free from CVD, antihypertensive pharmacotherapy, and relevant symptoms at the fifth checkup. Utilizing a threshold of ≥130/80 mmHg (systolic/diastolic) for elevated BP, participants were stratified into four phenotypes based on their BP at the fifth checkup (baseline in-office BP) and the weighted average of their BP from the first to the fifth checkups (historical BP): (1) non-hypertension; (2) new-onset hypertension (elevated baseline BP only); (3) potential hypertension (elevated historical BP only); and (4) persistent hypertension (elevated baseline and historical BP). Multivariable Cox proportional hazards regression was employed to evaluate the association between these BP phenotypes and incident CVD (a composite of myocardial infarction, heart failure, stroke, and atrial fibrillation).
Results
Over a median follow-up of 2.6 (IQR, 1.2–4.7) years, the incidence rates of CVD events (per 10,000 person-years) exhibited a stepwise increase across the BP categories (non-hypertension: 49.7; new-onset: 71.5; potential: 89.8; persistent: 135.0; log-rank P <0.001, Figure). In multivariable Cox regression models using non-hypertension as the reference, all hypertensive phenotypes were independently associated with an increased risk of CVD events. Notably, compared to new-onset hypertension (adjusted hazard ratio [aHR], 1.20; 95% CI, 1.13–1.28), both potential hypertension (aHR, 1.49; 95% CI, 1.35–1.64) and persistent hypertension (aHR, 1.97; 95% CI, 1.88–2.06) conferred substantially higher risks. This prognostic gradient was consistent across individual CVD endpoints and remained robust in multiple sensitivity analyses.
Conclusions
Categorizing BP based on both cross-sectional in-office readings and longitudinal historical averages is significantly associated with future CVD risk. These findings suggest that integrating historical BP trajectories with current clinical measurements enhances prognostic accuracy and facilitates superior risk stratification for CVD.FigureFor image description, please refer to the figure legend and surrounding text.