Population-Wide Assessment of Heart Rhythm and Physical Activity from 14-Day Recordings: The UK Biobank Cardiac Monitoring Study

The UK Biobank Cardiac Monitoring Study utilized 14-day patch-based ECG and accelerometer data from 27,658 participants to characterize population-wide heart rhythms and physical activity, revealing significant rates of undiagnosed atrial fibrillation and distinct circadian patterns of arrhythmias linked to activity levels.

The Gist

Imagine you have a car, and you only take it to the mechanic for a 15-minute test drive once every five years. The mechanic might tell you the engine sounds fine, but they'd miss the fact that the car sputters every time you hit a steep hill or the brakes squeal only when it's raining.

That is essentially what happened to heart health monitoring for decades. Doctors usually gave patients a 24-hour "holter monitor" (a small box that records heart rhythm) or a quick check-up. But many heart problems, like irregular heartbeats, are like that sputtering engine: they happen rarely, unpredictably, and often when the patient is asleep or resting.

The UK Biobank Cardiac Monitoring Study is like giving 28,000 people a high-tech, 14-day "black box" recorder for their hearts, along with a pedometer to track their movement, and then analyzing the data to see what's really going on under the hood.

Here is the breakdown of what they found, using simple analogies:

1. The Experiment: A Two-Week Vacation for the Heart

The researchers asked nearly 28,000 older adults (average age 71) to wear a small, sticky patch on their chest for 14 days straight.

• The Patch: Think of this as a tiny, invisible camera that never blinks. It recorded the heart's electrical signals (ECG) and how much the person was moving (accelerometry) every second of the day and night.
• The Scale: Previous studies were like looking at a single drop of water to understand the ocean. This study is the whole ocean. It is the largest study of its kind ever done.

2. The Big Discovery: The "Silent" Heart Problems

The most exciting finding is that short check-ups miss a lot.

• The "Hidden" Atrial Fibrillation (AF): AF is a dangerous irregular heartbeat that increases the risk of stroke. The study found that 3.2% of men and 1.7% of women had AF that no one knew about.
• The Time Factor: If you only watched these people for 24 hours, you would have missed 75% of the AF cases. It's like trying to spot a shooting star by looking at the sky for one minute; you have to watch for a long time to catch it. In fact, nearly a quarter of the new AF cases were only found in the second week of monitoring.
• The Pattern: These "silent" heart skips mostly happened at night when people were sleeping and their bodies were relaxed.

3. The "Day vs. Night" Heart Dance

The study also looked at how the heart behaves differently during the day compared to the night, using the movement data as a guide.

• The Night Owl Heart (AF): Irregular heartbeats (AF) loved the night. They peaked around 3:00 AM when people were asleep and inactive. It's as if the heart gets "lazy" and stumbles when the body is resting.
• The Daytime Sprinter (Ventricular Ectopy): On the flip side, other types of skipped beats (ventricular ectopy) happened mostly during the day, peaking around lunchtime when people were active and moving. It's like the heart getting a little "overheated" or stressed during the daily hustle and bustle.

4. The "Re-Test" Reliability

About 1,300 people wore the patch twice, with a gap of about five years between tests.

• The Result: The study found that a person's "heart personality" is surprisingly stable. If your heart rate and activity patterns were consistent on Day 1, they were likely consistent five years later. It's like a fingerprint; your daily rhythm is unique to you and stays that way for a long time.
• The Catch: While the rhythm of the day-to-day life was stable, the irregular heartbeats (the arrhythmias) were a bit more random. You might have a bad week of skipped beats one year and a clean week the next, making them harder to predict over the long haul.

5. Why This Matters

Think of this study as building a massive, detailed map of the heart's "weather patterns."

• Before: We only knew about the storms (heart attacks) that had already happened.
• Now: We can see the clouds gathering (subclinical arrhythmias) long before the storm hits.

Because this data is linked to the UK Biobank's other information (like brain scans, DNA, and medical history), scientists can now ask: "Does having a heart that skips a beat at night affect your memory five years later?" or "Do specific genes make your heart more likely to skip beats when you sleep?"

The Bottom Line

This study proves that to truly understand the heart, you can't just take a quick snapshot. You need a 14-day movie. By watching people for two weeks, the researchers found thousands of hidden heart problems that would have otherwise gone unnoticed. This opens the door to catching diseases earlier, preventing strokes, and understanding how our daily lives (sleep, activity, age) shape the rhythm of our hearts.

Note: This paper is a "preprint," meaning it is a fresh report that hasn't been through the final peer-review "quality check" yet, but it offers a fascinating glimpse into the future of heart health monitoring.

Technical Summary

1. Problem Statement

While wearable sensors have revolutionized cardiovascular monitoring, significant gaps remain in leveraging this data for large-scale disease prediction and understanding the natural history of subclinical arrhythmias.

• Limitations of Current Data: Existing studies often suffer from small sample sizes, narrow scopes of arrhythmia types, and reliance on patient-based cohorts rather than general populations.
• Data Integration Gaps: There is a scarcity of prospective datasets that integrate long-term wearable sensor data (ECG and accelerometry) with complementary clinical measurements, longitudinal health outcomes, multi-omics, and imaging.
• Clinical Need: Short-term monitoring often fails to detect paroxysmal (intermittent) arrhythmias like Atrial Fibrillation (AF), which are clinically silent but carry high risks of stroke and mortality. There is a need to characterize the circadian patterns of arrhythmias and their relationship with physical activity in an older adult population.

2. Methodology

Study Population & Design

• Cohort: A subset of the UK Biobank imaging study participants (aged 40–69 at recruitment, now older).
• Phases:
  • Pilot Phase (2015–2018): Used Zio XT monitors (iRhythm). Initially 200 participants, expanded to 7,795 participants aged ≥60 without prior AF.
  • Main Phase (2019–Ongoing): Used BodyGuardian MINI monitors (Preventice/Boston Scientific). Targeted participants aged ≥65 (expanded to ≥60 during the pandemic). Included individuals with prior AF from May 2022.
• Protocol: Participants wore patch monitors for 14 days. Devices recorded continuous single-lead ECG and triaxial accelerometry.
• Sample Size: 27,658 participants analyzed (7,795 pilot; 21,141 main). 1,353 participants had repeat assessments (median interval 5 years).

Data Acquisition & Processing

• ECG Analysis:
  • Main Phase: Processed by Preventice's partner (MDT) using automated algorithms (ECGLAB) and expert manual review. Defined arrhythmias included AF, SVT, VT, and conduction blocks. "New AF" was defined as episodes >30s in participants with no prior self-reported AF.
  • Pilot Phase: Processed by iRhythm using FDA-approved algorithms (summary data only).
  • Custom Processing: In-house deep learning-based QRS detection algorithms were applied to derive beat-to-beat RR intervals for circadian analysis. Non-wear time was imputed using mean RR intervals from other days.
• Accelerometer Processing:
  • Data harmonized using device-specific pipelines to account for sampling frequency differences (Zio XT: ~1.56 Hz vs. BodyGuardian: 25 Hz).
  • Signals calibrated using the Actipy Python package to correct sensor bias.
  • Movement derived as the Euclidean norm of calibrated axes; summarized in 1-minute epochs.
• Inclusion Criteria: Participants required ≥3 days of analyzable ECG data and coverage in every hour of the 24-hour cycle.

Statistical Analysis

• Assessed wear time, arrhythmia prevalence, and circadian patterns.
• Used Wilcoxon–Mann–Whitney tests for continuous data and Chi-squared/Fisher's exact tests for categorical data.
• Calculated Intraclass Correlation Coefficients (ICC) and Pearson correlations to assess repeatability in the subset with repeat recordings.

3. Key Contributions

• Scale: The largest population-based dataset of prolonged ambulatory ECG monitoring to date (~28,000 participants), an order of magnitude larger than previous cohorts (e.g., MESA, ARIC).
• Dual-Modality Integration: Uniquely combines high-resolution 14-day ECG with calibrated accelerometry, allowing for the correlation of arrhythmia burden with physical activity and circadian rhythms.
• Longitudinal Repeatability: Includes a unique subset of >1,300 participants with repeat 14-day recordings separated by ~5 years, enabling the assessment of the long-term stability of cardiac phenotypes.
• Open Data Pipeline: Establishes a robust pipeline for deriving RR intervals and activity metrics from raw wearable data, facilitating future multi-omics and imaging linkage.

4. Key Results

Demographics & Compliance

• Cohort: Mean age ~71 years; 49.9% women.
• Wear Time: High compliance. Median wear time was 13.2 days (Main) and 14.0 days (Pilot). 92.6% of main phase recordings exceeded 7 days.

Arrhythmia Prevalence (Main Phase)

• Atrial Fibrillation (AF): Detected in 2.4% of participants (undiagnosed).
  • Demographics: More common in men (3.2% vs 1.7%) and those >70 years.
  • Nature: 68% were paroxysmal. Median burden 2.4%.
  • Detection Yield: Only 25.3% detected in the first 24 hours; 74.3% detected by day 7; 25.7% detected only in the second week, highlighting the necessity of long-term monitoring.
• Supraventricular Tachycardia (SVT): 2.6% prevalence. 90.3% of episodes lasted <1 minute.
• Ventricular Tachycardia (VT): 12.1% prevalence (higher in men and older adults). 99.6% were non-sustained (<30s); sustained VT was rare (0.4%).
• Conduction Disorders: First-degree AV block (20.9%), Second-degree (1.7%), Third-degree (<0.1%).

Circadian Patterns & Activity

• AF Timing: Peaked during nocturnal inactivity (around 3:00 AM) and reached a nadir in the late morning (11:00 AM).
• Ventricular Ectopy (VE): Showed the opposite pattern, increasing with daytime activity and peaking around 1:00 PM.
• Heart Rate/Activity: Clear 24-hour rhythms with peaks in the morning (09:00–10:00), mid-afternoon, and early evening. Women had higher mean heart rates than men; older adults had lower rates.

Repeatability

• Physiological Traits: Strong reproducibility for mean heart rate and acceleration (ICC ~0.85 for HR, 0.69 for acceleration) across 5-year intervals.
• Arrhythmia Burden: Moderate reproducibility for AF (ICC 0.58) and lower for ectopy (ICC ~0.37–0.38), likely due to the episodic nature of these events.

5. Significance

• Clinical Impact: Demonstrates that short-term monitoring significantly underestimates paroxysmal arrhythmia prevalence. The finding that ~25% of AF cases are detected only in the second week of monitoring supports the shift toward extended monitoring in clinical guidelines.
• Pathophysiological Insight: Confirms distinct circadian mechanisms: AF is associated with high vagal tone (night/inactivity), while ventricular ectopy is linked to sympathetic drive and physical exertion.
• Future Research: This dataset serves as a foundational resource for the UK Biobank. Linking these rhythm and activity phenotypes with brain imaging, cardiac MRI, and multi-omics will allow researchers to:
  • Investigate the natural history of asymptomatic rhythm disturbances.
  • Determine causal links between subclinical arrhythmias and neurodegenerative diseases (e.g., dementia, stroke).
  • Develop better risk prediction models for cardiovascular events.
• Feasibility: Proves that 14-day patch monitoring is feasible and well-tolerated in large-scale epidemiological studies involving older adults.