Circadian Blood Pressure Variations Computed From 1.7 Million Measurements in an Acute Hospital Setting

Adam Mahdi, Peter Watkinson, Richard J. McManus, Lionel TarassenkoAmerican 2019 · DOI

TL;DR

This study analyzed over 1.7 million blood pressure (BP) measurements from hospitalized patients, revealing that typical "dipping" BP patterns during sleep are largely absent in older adults, replaced by a late-night rise. This finding, observed even in patients without a hypertension diagnosis, suggests that hospital BP monitoring can reveal important cardiovascular risk patterns that mirror community observations.

What Problem Does This Paper Address?

Understanding the natural, daily fluctuations of blood pressure (BP), known as circadian rhythms, is crucial for cardiovascular health. In the community, BP typically drops during sleep (a "dipping" pattern) and rises during the day. Deviations from this pattern, particularly a lack of dipping or even a rise at night, are linked to increased cardiovascular risk. However, knowledge of these circadian BP variations in the acute hospital setting, where patients are often monitored closely, has been limited. This gap is significant because hospital stays provide a unique opportunity to capture detailed, real-time BP data, potentially revealing risk factors that might be missed in sporadic clinic visits or even standard 24-hour ambulatory monitoring (ABPM) outside of a hospital context. The paper aims to characterize these in-hospital circadian BP patterns and explore how they vary with age, sex, and hypertension status.

Methods

The researchers conducted a retrospective analysis of BP data collected from adult general wards in four acute hospitals in Oxford, UK, between March 2014 and April 2018. They utilized a standardized electronic health record system that captures vital signs, including BP, and calculates an Early Warning Score (EWS) to flag physiological deterioration.

The study design involved:

  • Data Extraction: Pulling BP data along with demographic information (age, sex) and admission details from the electronic health record system.
  • Patient Selection: Including patients aged 16 and older who had at least three BP observations, with at least one during the defined nighttime (midnight to 5:59 AM) and daytime (10:00 AM to 7:59 PM) periods, and with observations at least 24 hours apart. Admissions with no nighttime or daytime observations were excluded.
  • Data Aggregation: For each patient, hourly averages of systolic and diastolic BP were calculated across their entire hospital stay, creating a 24-hour profile. This approach ensured that patients with varying lengths of stay contributed equally to the analysis.
  • Stratification: The analysis was stratified by age (in decades) and sex. Additional analyses explored the impact of hypertension diagnosis and treatment, as well as physiological instability (indicated by EWS scores).

Datasets

The study utilized a large, real-world dataset comprising:

  • Source: Electronic health records from four acute hospitals in Oxford, UK.
  • Timeframe: March 2014 to April 2018.
  • Scale: Data from 41,455 unique patient admissions, resulting in approximately 1.7 million sets of vital-sign measurements.
  • Key Data Points: Systolic and diastolic blood pressure, heart rate, and other vital signs used for EWS calculation. Linked prescription data and ICD-10 codes for hypertension were also used.

This dataset is notable for its size and the context of acute hospital care, offering a different perspective than community-based ABPM studies.

Key Insights and Ideas

  • Age-Dependent BP Patterns: The most striking finding is the stark difference in circadian BP patterns between younger and older hospitalized patients. Younger individuals largely exhibit the expected "dipping" pattern, while older individuals (over 50) show a pronounced late-night rise in systolic BP.
  • Hypertension Status Irrelevant for Older Adults' Nighttime Rise: This late-night BP rise in older patients was observed regardless of whether they had a pre-existing hypertension diagnosis or were on medication. This suggests that the hospital environment or age-related physiological changes, rather than just diagnosed hypertension, play a significant role in altering these patterns.
  • Hospital Data Mirrors Community Findings: Despite the unique hospital environment, the observed circadian BP patterns in hospitalized patients closely resemble those seen in community-dwelling populations. This validates the use of hospital data for studying these rhythms and suggests that hospital monitoring can capture clinically relevant cardiovascular information.
  • Potential for Risk Stratification: The ability to identify these deviations from typical BP patterns in a large hospitalized cohort opens doors for opportunistic or systematic screening to identify patients at higher cardiovascular risk.

Key Results

  • Dataset Size: 41,455 unique patient admissions, with 1.7 million BP measurements.
  • Typical Dipping: Observed only in younger age groups (up to 40-49 for men, 30-39 for women).
  • Late Nocturnal Rise: Present in older age groups (above 50), with amplitude increasing with age. This rise occurred even in patients without a prior hypertension diagnosis.
  • Hypertension Cohorts (60+): Both hypertensive and normotensive elderly patients showed similar late nocturnal BP rises.
  • Heart Rate: Across all age groups, heart rate decreased during the night, reaching its lowest point before awakening, which the authors argue suggests the BP rise is not solely due to arousal from sleep.
  • Physiological Instability: Removing data from physiologically unstable patients had little effect on the observed BP patterns, suggesting the findings are robust.

Limitations and Future Work

  • Retrospective Nature: The study is retrospective, relying on existing data collection protocols.
  • Selection Bias: Patients with more severe conditions or those requiring more frequent monitoring might be overrepresented, potentially influencing the data. However, the authors attempted to mitigate this by generating a single 24-hour profile per patient.
  • Measurement Method: The study used standard ward BP measurements, not ABPM. The authors acknowledge that the method of taking BP (e.g., manual cuff application) might influence results, potentially causing sleep disruption and sympathetic activation, though they present evidence against this being the sole cause of the observed patterns.
  • Hospital Environment Influence: External factors like ward routines and admission times might synchronize patient rhythms, potentially influencing the observed patterns.
  • Future Work: The authors suggest exploring the effects of different admission times, investigating comorbidities, and determining whether these in-hospital observed patterns persist after discharge and correlate with long-term cardiovascular risk. They also propose investigating the mechanistic link between BP measurement practices and potential sleep disruption.

Summary updated

Abstract

Abstract BACKGROUND Knowledge of the circadian blood pressure (BP) variations in the acute hospital setting is very limited. METHODS This is a retrospective analysis of BP data for in-hospital patients stratified by age and sex. We used data collected with the help of a standardized electronic health record system between March 2014 and April 2018 on the adult general wards in 4 acute hospitals in Oxford, UK. RESULTS A total of 41,455 unique patient admissions with 1.7 million sets of vital-sign measurements have been included in the study. The typical 24-hour systolic BP profile (dipping pattern during sleep followed by a gradual increase during the day) was only seen in the younger age groups (up to 40–49 for men and 30–39 for women). For older age groups, there was a late nocturnal rise in systolic BP, the amplitude of which increased with age. The late nocturnal BP rise above the age of 50 was seen whether or not the patient was treated for or previously identified with hypertension. CONCLUSION Hospitalized patients’ circadian patterns of BP largely mirror those found in the community. High-quality hospital data may allow for the identification of patients at significant cardiovascular risk through either opportunistic screening or systematic screening.
BibTeX
@inproceedings{mahdi2019circadian,
  author = {Adam Mahdi and Peter Watkinson and Richard J. McManus and Lionel Tarassenko},
  title = {{Circadian Blood Pressure Variations Computed From 1.7 Million Measurements in an Acute Hospital Setting}},
  booktitle = {American},
  year = {2019},
  doi = {10.1093/ajh/hpz130},
}
Figure