The minimum number of blood pressure measurements needed and thresholds for visit-to-visit blood pressure variability to predict cardiovascular disease in primary care patients

This study demonstrates that at least five blood pressure measurements are sufficient to calculate visit-to-visit variability, identifying specific standard deviation, coefficient of variation, and average real variability thresholds that effectively predict cardiovascular disease risk in primary care patients.

The Gist

Imagine your blood pressure isn't just a single number you get when you visit the doctor, like a snapshot of a photo. Instead, think of it as a movie. Sometimes the movie is smooth and steady; other times, it's full of shaky camera work, sudden zooms, and wild jumps.

This study is all about figuring out how much "shaky camera work" (variability) in your blood pressure movie is dangerous, and how many scenes (measurements) we need to watch to know if the movie is going to end in a disaster (heart disease).

Here is the breakdown of the research in simple terms:

1. The Problem: The "Shaky" Blood Pressure

We all know that high blood pressure is bad for your heart. But doctors usually just look at one or two numbers and say, "Okay, that's high, take a pill." They often ignore the variability.

Think of your blood pressure like a rollercoaster.

• Steady High: A rollercoaster that stays high but flat is dangerous.
• The Rollercoaster Ride: A rollercoaster that goes up and down wildly, even if the average height is okay, is also dangerous.

This study found that if your blood pressure jumps around a lot from one doctor's visit to the next (even if the average isn't super high), it's a huge warning sign for future heart attacks and strokes.

2. The First Question: How Many Measurements Do We Need?

To know if your blood pressure is "shaky," you need to measure it several times. But how many times?

• The Old Idea: Maybe we need 8 or 10 visits to be sure.
• The New Finding: The researchers did a math test and found that 5 visits are enough.

The Analogy: Imagine trying to guess the weather pattern of a week. If you check the temperature 3 times, you might get lucky or unlucky. But if you check it 5 times, you have a pretty good idea of whether it's going to be stormy or sunny. You don't need to wait for 8 days to know the pattern.

The Result: You need at least 5 blood pressure readings taken on different days to get a reliable "variability score."

3. The Second Question: What is the "Danger Zone"?

Once we know we need 5 readings, the researchers asked: "How much jumping around is too much?" They set up a "speed limit" for your blood pressure variability.

They used three different ways to measure the "shakiness" (Standard Deviation, Coefficient of Variation, and Average Real Variability). Think of these as three different types of speedometers.

The Speed Limits (The Cut-offs):
If your blood pressure jumps more than these amounts from visit to visit, your risk of heart disease goes up significantly:

• For Systolic (The top number): If it jumps more than 19 mmHg (about 19 units) between visits, you are in the danger zone.
• For Diastolic (The bottom number): If it jumps more than 11 mmHg, you are in the danger zone.

The Metaphor: Imagine your blood pressure is a car driving down a highway.

• If you are driving at a steady 60 mph, that's fine.
• If you are driving at a steady 80 mph, that's bad (high pressure).
• But if you are driving at an average of 60 mph, but you are slamming on the brakes and hitting the gas pedal wildly, swinging between 40 and 80 mph every few minutes? That is the danger. This study says that wild swinging is just as dangerous as driving too fast.

4. Who is Most at Risk?

The study found some surprising things:

• The "Normal" People: Even people whose average blood pressure is considered "normal" (under 140) are at risk if their numbers jump around wildly. It's like a calm-looking lake that has hidden, violent underwater currents.
• Women: The study noticed that women with high variability seemed to have a steeper drop in survival rates compared to men. The researchers suspect this might be related to hormonal changes around menopause (around age 55), which might make the blood vessels more sensitive to these jumps.

5. What Does This Mean for You?

This research is a wake-up call for how we check our health.

• Don't just check once: If you go to the doctor and get one reading, it's not enough to judge your heart risk.
• Keep the records: Your doctor needs to see your history. If you have at least 5 different records of your blood pressure over time, they can calculate your "variability score."
• The New Warning Sign: If your numbers are jumping around a lot (more than 19 points for the top number), your doctor might need to treat you for heart disease risk, even if your average blood pressure looks okay.

The Bottom Line

Your blood pressure shouldn't be a rollercoaster. This study tells us that 5 measurements are enough to see if your blood pressure is too "shaky," and if it jumps more than 19 points between visits, it's a major red flag for your heart. It's a simple, new way to catch heart trouble before it happens.

Technical Summary

1. Problem Statement

Visit-to-visit blood pressure variability (VVV BPV) is an emerging risk factor for cardiovascular disease (CVD), yet it remains underutilized in clinical practice. While previous research has established a link between BPV and CVD, significant gaps exist:

• Lack of Clinical Thresholds: There are no definitive, clinically relevant cut-off values for VVV BPV (measured via Standard Deviation [SD], Coefficient of Variation [CV], or Average Real Variability [ARV]) derived from real-world office blood pressure readings to predict CVD events.
• Measurement Uncertainty: It is unclear what the minimum number of blood pressure (BP) measurements required is to reliably calculate BPV for risk prediction. Most existing studies rely on controlled clinical trials or small sample sizes, limiting generalizability to primary care settings.
• Population Gap: Previous studies often focused on older populations (>60 years) or those with pre-existing CVD, leaving a gap in understanding VVV BPV in younger, primary care populations (ages 18–55) before the onset of disease.

2. Methodology

The study utilized a two-part design using data from the electronic practice-based research network (ePBRN) in Southwestern Sydney, Australia (2006–2019).

Data Source & Cohort Selection:

• Data: Routinely collected Electronic Health Records (EHR) from General Practitioner (GP) clinics and linked hospital data.
• Study 1 (Measurement Validation): Included 1,549 patients aged 18–55 with at least 8 BP readings prior to age 55. The goal was to determine the minimum number of readings needed.
• Study 2 (Threshold Derivation): Included 3,022 patients aged 18–55 with at least 5 BP readings.
• Inclusion Criteria: No prior diagnosis of atherosclerotic CVD or atrial fibrillation at the index date (age 55).
• Exclusion Criteria: Missing sex/year of birth, no follow-up after age 55, or biologically implausible BP values (SBP <60 or >250 mmHg; DBP <40 or >140 mmHg).

Statistical Analysis:

• Agreement Analysis: Compared SD calculated from 3–7 readings against a reference of 8 readings using Pearson's correlation and Intraclass Correlation Coefficient (ICC).
• Risk Modeling: Used Cox proportional hazards models to estimate Hazard Ratios (HR) for the first CVD event (fatal/non-fatal CHD, stroke, TIA, heart failure, AF, PAD) occurring after age 55.
• Adjustments: Five models were tested, progressively adjusting for sex, mean BP, comorbidities, laboratory values (cholesterol, eGFR, glucose, BMI), and medications.
• Cut-off Determination: Percentile-based cut-offs (10th–90th) were derived. The optimal cut-off was identified based on the highest Concordance Index (C-statistic) and time-varying Area Under the Curve (AUC) over 2, 4, and 5-year horizons.
• Handling Missing Data: Multiple Imputation by Chained Equations (MICE) was used to generate 20 imputed datasets for missing covariates.

3. Key Contributions

• Minimum Measurement Standard: Established that five (5) BP measurements are sufficient to estimate VVV BPV with good reliability (ICC 0.79) compared to a reference of eight measurements.
• Clinically Actionable Thresholds: Defined specific cut-off values for VVV BPV metrics (SD, CV, ARV) that predict CVD risk in a primary care setting.
• Real-World Evidence: Provided the first large-scale analysis of VVV BPV cut-offs using routine EHR data from a general practice population, rather than controlled clinical trial data.
• Subgroup Insights: Identified that high BPV poses a significant risk even in patients with mean systolic BP <140 mmHg and highlighted a gender-specific risk profile where women with high BPV face steeper survival declines than men.

4. Key Results

A. Minimum Number of Measurements:

• Comparing 5 readings to 8 readings yielded a correlation coefficient of 0.80 and an ICC of 0.79 (indicating "good" reliability).
• Fewer than 5 readings showed poor to moderate agreement.

B. Predictive Association (Hazard Ratios):

• Higher VVV BPV was significantly associated with increased CVD risk across all metrics (SD, CV, ARV) and both systolic (SBP) and diastolic (DBP) pressures, even after full adjustment (Model 5).
• Systolic BPV (per 1-SD increase):
  • SD: HR = 1.71 (95% CI: 1.49–1.95)
  • CV: HR = 1.69 (95% CI: 1.49–1.92)
  • ARV: HR = 1.36 (95% CI: 1.22–1.51)
• Diastolic BPV (per 1-SD increase):
  • SD: HR = 1.95 (95% CI: 1.72–2.21)
  • CV: HR = 1.90 (95% CI: 1.69–2.13)
  • ARV: HR = 1.55 (95% CI: 1.39–1.72)
• Note: SD consistently demonstrated slightly higher effect estimates than CV and ARV.

C. Optimal Cut-off Values:
The study identified the following thresholds for predicting CVD risk:

• Systolic BP:
  • SD: 19 mmHg
  • CV: 14%
  • ARV: 15 mmHg
• Diastolic BP:
  • SD: 11 mmHg
  • CV: 12%
  • ARV: 11 mmHg

D. Predictive Performance:

• The C-statistic (AUC) for 5-year prediction was approximately 0.74 for SBP-SD and 0.73 for DBP-SD, indicating moderate discrimination.
• Performance was consistent across 2, 4, and 5-year time horizons.

E. Subgroup Findings:

• Mean BP <140 mmHg: Patients with normal mean BP but high variability had a higher hazard ratio for CVD compared to those with high mean BP.
• Gender: Among patients with high BPV (SD ≥19 mmHg), women showed a steeper decline in survival probability compared to men, potentially linked to perimenopausal hormonal changes.

5. Significance and Implications

• Clinical Practice: The findings suggest that routine BP monitoring in primary care should aim for at least five separate visits to reliably calculate BPV. Clinicians should use the identified cut-offs (e.g., SBP SD >19 mmHg) to identify high-risk patients who might otherwise be missed by mean BP alone.
• Risk Stratification: VVV BPV can serve as a tool to stratify "residual risk" in patients who have achieved target mean BP levels, particularly those under 55.
• Policy: The study advocates for policy changes to fund regular BP screening (e.g., expanding Medicare item numbers for chronic disease and menopause assessments) to ensure continuity of care and data collection for BPV calculation.
• Future Research: The authors recommend integrating VVV BPV into CVD risk calculators and conducting further validation in younger cohorts and diverse populations.

Limitations Noted:

• The study relied on EHR data with variable measurement intervals, making ARV (which requires regular intervals) slightly less predictive than SD or CV.
• The cohort was restricted to patients with sufficient data (≥5 readings), which may introduce selection bias toward patients with more chronic conditions or better healthcare engagement.
• The study focused on patients aged 55+ for outcomes, potentially underestimating risk in younger individuals with early-onset CVD.