BP/AF MODE Study

Abstract

Atrial fibrillation is a highly prevalent and incidental arrhythmia, often asymptomatic, and frequently detected incidentally or in association with a stroke. Subclinical atrial fibrillation increases cardioembolic risk, highlighting the need for timely diagnosis.

New wireless devices capable of recording heart rhythm, combined with innovative artificial intelligence tools, could be useful in the prediction and detection of this arrhythmia.

Objective: To determine the usefulness of home blood pressure and heart rhythm monitoring strategy in the detection of subclinical atrial fibrillation.

Methods: Observational, cohort, prospective, multicenter study involving 25 researchers from six Latin American countries. Home blood pressure monitoring and single-lead electrocardiogram recording were performed in a population at moderate to high risk of developing atrial fibrillation. A minimum of twenty 30-second electrocardiographic and blood pressure recordings over 7 days using an Omron Complete Hem-7530 T ECG device will be uploaded from a mobile phone app and then sent to a database for analysis.

Conclusion: The results of this study can provide a simple and accessible home monitoring system for detecting subclinical atrial fibrillation and for optimizing the predictive capacity of arrhythmia risk scores through deep learning.

Key words: hypertension, atrial fibrillation, home monitoring, artificial intelligence

Introduction

Atrial fibrillation (AF) is a very common arrhythmia, with a prevalence, depending on the subject's gender, ranging from 3% to 5% in people aged 70 to 74 years and 7% to 10% in those aged 80 to 84 years. It is estimated that by 2050, one-sixth of the world's population will have AF ¹.

This arrhythmia is often asymptomatic, detected incidentally during intraoperative monitoring or in patients with implanted or wireless devices, or as the first manifestation of a stroke; furthermore, in approximately 5% of cases, AF is detected within a year of the event ².

It is estimated that 4% of hypertensive patients are carriers of AF, while during the evolution of the disease another 4% will develop an episode of AF ³. Subclinical AF (sAF) generally implies a lack of timely diagnosis, establishing the real possibility of the occurrence of cardioembolic stroke and peripheral arterial embolisms. Although the incidence is lower than that of clinical AF, it can generate consequences that are no less worrying in terms of years of life lost or quality of life ⁴.

There are several risk scores that allow the identification of AF at-risk population, and as stated, the technological capabilities of implanted devices allow for the detection of these asymptomatic arrhythmias.

In the last decade, there has been a significant development of wireless devices for heart rhythm monitoring. This advancement, along with the emergence of tools based on machine learning and deep learning, has sparked growing interest in the scientific community, promoting the incorporation of these emerging technologies into patient monitoring and care ^{5 6}.

In this context, we propose to evaluate the usefulness of the implementation of a program based on obtaining 1-lead electrocardiographic recordings in patients at high risk of developing AF may have in the detection of asymptomatic arrhythmias.

Objectives

The primary objective of this study is to determine the usefulness of a home blood pressure and heart rate monitoring strategy in detecting AF in hypertensive patients in Latin America.

The secondary objective is to identify predictors of AF in a sample of Latin American hypertensive subjects at moderate to high risk of developing this arrhythmia, thereby optimizing the predictive capacity of AF risk scores.

Materials and Methods

The BP/AF Mode study is a multicenter, observational, prospective, cohort trial involving 25 researchers from six countries: Argentina, Chile, Colombia, Ecuador, Mexico, and Peru. The study was submitted to the IASC Academia (Inter-American Society of Cardiology) and approved under Project File Number 052, dated January 29, 2025. It will be conducted in full compliance with the World Medical Association Codes of Ethics for experiments involving human subjects.

Inclusion criteria

1. Hypertensive patients of both sexes over 75 years of age.
2. Hypertensive patients between 65 and 75 years of age are associated with one of the following risk factors:
1. Obesity
2. Type 2 diabetes
3. Obstructive sleep apnea syndrome
4. Electrocardiogram showing left ventricular hypertrophy
5. Left ventricular hypertrophy on echocardiogram
6. Left atrial dilation
7. Coronary artery disease
8. Chronic kidney disease
9. Heart failure with ejection fraction > 50%
10. Alcohol intake
11. Frequent supraventricular ectopic beats

Exclusion criteria

1. History of any type of atrial fibrillation.
2. Receiving pharmacological treatments with potential impact on heart rhythm.
3. History of potentially malignant ventricular arrhythmias.
4. Uncontrolled hypothyroidism or hyperthyroidism.
5. Significant chronic obstructive pulmonary disease.

The diagnostic criteria for the variables that make up the inclusion criteria are detailed below:

• Definition of hypertension: patient with office BP > 140-90 mmHg and > 130-80 mmHg on 24-hour ABPM or > 135-85 mmHg on daytime ABPM, or who is receiving treatment with antihypertensive drugs⁷.
• Definition of obesity: body mass index ≥ 30 kg/m2 or waist circumference > 88 cm in women and > 102 cm in men.
• Definition of type 2 diabetes: glycated hemoglobin > 6.5%, fasting blood glucose > 126 mg/dL, or a 2-hour glucose tolerance test > 200 mg/dL, or typical symptoms of hyperglycemia associated with a random blood glucose level > 200 mg/dL, or treatment with oral hypoglycemic agents or insulin⁸.
• Definition of Obstructive Sleep Apnea Syndrome: polysomnography with more than 5 apnea/hypopnea episodes per hour or more than 1.4 episodes per hour of desaturation greater than 4%⁹.
• Definition of electrocardiogram with left ventricular hypertrophy: strain patent, Sokolow-Lyon voltage criteria, Cornell, or Cornell product.
• Definition of left ventricular hypertrophy on echocardiogram: ventricular mass index > 95 kg/m2 in women and 115 kg/m2 in men¹⁰.
• Definition of left atrial enlargement: indexed left atrial volume on echocardiogram > 34 ml/m2⁸.
• Definition of coronary artery disease: history of acute myocardial infarction, hospitalization for acute coronary syndrome, or myocardial revascularization.
• Definition of chronic kidney disease: estimated glomerular filtration rate <60 ml/min/1.75 m2.
• Definition of heart failure with ejection fraction >50%: clinical syndrome of heart failure with an H2FPEF score >4 points, and ejection fraction >50% by volumetric echocardiographic method, magnetic resonance imaging, or radiocardiogram¹¹.
• Definition of cerebrovascular disease: history of paresis, paralysis, or amaurosis fugax lasting more than 30 minutes, documented in clinical history or imaging studies with cerebral, cerebellar, or brainstem infarction or hemorrhage.
• Definition of peripheral vascular disease: history of amputation due to critical ischemia, intermittent claudication associated with stenosis greater than 50% in at least one territory, or an ankle-brachial index less than 0.9 in the affected lower limb.
• Definition of alcoholic beverage intake: intake of more than four 75 ml glasses of wine or equivalent per day.
• Definition of frequent supraventricular ectopic beats: more than 10 supraventricular ectopic beats per hour on a 24-hour Holter electrocardiogram.

Cardiovascular risk stratification

The HEARTS cardiovascular risk estimator will be used, which includes gender, age, history of ischemic heart disease, cerebrovascular disease, or peripheral vascular disease, chronic kidney disease, diabetes mellitus, smoking, total cholesterol, and systolic blood pressure. Patients with a history of ischemic heart disease, cerebrovascular disease, or peripheral vascular disease do not require stratification, as they are considered very high risk, as are subjects with chronic kidney disease or diabetes mellitus, who are considered high risk¹².

High to moderate risk of developing AF

A HARMS2-AF score > 5 or a modified Taiwan AF score > 4 will be considered moderate to high risk of developing AF^{13 14}.

Frailty Assessment

Frailty will be assessed using the FRAIL index. It includes five domains: weight loss, exhaustion, weakness, slow gait, and reduced physical activity. The score ranges from 0 to 5, with 0 being non-frail and 5 being extremely frail. Subjects with a score of 0 were classified as "non-frail," 1-2 as "pre-frail," and ≥3 as "frail"¹⁵.

Procedures

The 30-second home ECG tracings will be obtained using the Omron Complete Hem-7530 T ECG device (Omron Healthcare Co., Ltd., Japan). The connection and smartphone app download will be available from Omron Connect free of charge on the Google Play Store: https://play.google.com/store/apps/details?id=jp.co.omron.healthcare.omron_connect&hl=es_419&pli=1 and on the iOS App Store: https://apps.apple.com/mx/app/omron-connect/id1003177043. Two duplicate ECG recordings will be required daily: the first in the morning before breakfast and taking the prescribed medication, and another in the evening before dinner and taking the afternoon/evening medication. Additional ECG recordings will be required if the patient experiences any symptoms. The study investigator will work with the patient to download the app and perform the first electrocardiogram tracing and transmit it to the database. The patient will be given instructions on how to correctly perform the ECG and transmit it to the platform, along with a patient diary. Simultaneously with obtaining the ECG tracings, the patient will take two BP measurements, in a seated position, separated by one minute, following the recommendations of the Latin American Society of Hypertension guidelines5. After 1 week, the patient will return the device to the research center. A recording will be considered valid when at least 80% of the ECG tracings and BP measurements are valid.

Statistical analysis

Considering those patients with a modified Taiwan AF score > 5 points, whose annual incidence of atrial fibrillation ranges between 1.54% and 6.98%, or a HARMS2-AF score > 7 points, whose annual incidence of atrial fibrillation ranges between 1.35% and 6.62%, it will be required to include 499 patients to achieve a 5% margin of error, with a 99% confidence level. If the sample calculation does not achieve the expected results, recruitment will continue until 25 cases of atrial fibrillation are reached.

Continuous variables will be reported as means with their standard deviations for normally distributed variables, or median and interquartile range for non-normally distributed variables. Discrete variables will be reported as absolute values and percentages. Unpaired ANOVA will be used to analyze normally distributed variables, while the Kruskal-Wallis test will be used for non-normally distributed variables. Differences in proportions will be assessed using the Chi-square test. A two-tailed p-value < 0.05 will be considered statistically significant.

Discussion

Without any doubt, healthcare is facing a paradigm shift in patient care as a result of the shift toward the digital era, driven by technological advances already available to the global community. Cardiovascular diseases are not immune to these changes, forcing healthcare teams to incorporate new wireless devices through programs that include patient training and, therefore, foster their empowerment.

A first question regarding AF detection is which patient population can benefit most, and it is clear that these will be those at greatest risk.

Although hypertension is perhaps the most important risk factor for developing AF, it is not the only one. The HARMS2-AF risk score for developing a new arrhythmia includes age over 60 years, obesity, male gender, the presence of obstructive sleep apnea syndrome, smoking, and consuming at least seven standard servings of alcoholic beverages per week10. Individuals with a HARMS2-AF score of between 5 and 9 points increase their risk of developing a new AF by 12.79 times (95% CI: 8.93–18.33), and with a score of 10 to 14 points, the risk increases by 38.70 times (95% CI: 26.96–55.54). The area under the curve for the score was 0.782, higher than the Framingham-AF model of 0.568.

The modified Taiwan AF score considered age, male sex, the presence of hypertension, heart failure, coronary artery disease and chronic kidney disease, and excluded alcoholism from the original score in the risk stratification of AF. The group of patients with a score of 1 or 2 points had an annual incidence of arrhythmia of 0.2%, in those subjects with a score between 4 and 9 points the incidence was 1.33% per year, while in those with a score of 9 to 14 points the incidence increased to 3.6% per year11. The area under the curve to detect new atrial fibrillation was 0.861 in the first year of follow-up, progressively decreasing to 0.751 at 16 years of follow-up.

In this scenario, it is assumed that the use of algorithm platforms may improve risk characterization, especially when applied to a specific population, which may be different from those used in the literature to define prevalence and incidence rates.

A second aspect to consider is whether low-complexity equipment can provide important information for identifying sAF.

The RITMO study was a program to optimize the detection of paroxysmal AF. Subjects over 65 years of age with hypertension or heart failure and no symptoms compatible with arrhythmias were included. AF risk was stratified using the Stroke Risk Analysis (SRA) using a 1-hour electrocardiogram strip. Patients classified as high risk were prescribed a 7-day home monitoring protocol with the Kardia 6 device (OMRON, AliveCor®). Two-point eight percent of patients with hypertension were detected with AF using the SRA, and 10.2% of new cases were detected during home electrocardiogram monitoring¹⁶.

In a project that included diabetic patients in primary prevention, an automatic device was used to measure blood pressure and detect AF, Microlife WatchBP Office AFIB. The prevalence of AF was 1.9% (95% CI 1.3–2.6), increasing from 0.5% in patients younger than 65 years to 2.2% in patients between 65 and 74 years, and to 4.3% in those older than 75 years. The authors concluded that AF screening with this type of device can be considered a standard procedure in this type of clinical situation¹⁷.

The BP/AF MODE study proposes an automated blood pressure monitor with the ability to record electrocardiograms in one lead for 30 seconds. The study population is at moderate to high risk for AF.

Conclusions

The results of this study can provide a simple and accessible home monitoring system for detecting subclinical atrial fibrillation and optimizing the predictive capacity of arrhythmia risk scores through deep learning.

The usefulness of a blood pressure and heart rate monitoring system, with patient or caregiver participation and low-cost equipment, to detect AF cases will be established.

Deep learning tools could improve risk profile definition beyond the capabilities of the scores used so far.

Bibliography

¹Miyasaka Y, Barnes ME, Gersh BJ, et al. Secular trends in incidence of atrial fibrillation in Olmsted County, Minnesota, 1980 to 2000, and implications on the projections for future prevalence. Circulation 2006; 114: 119 - 125. doi: 10.1161/CIRCULATIONAHA.105.595140. Erratum in: Circulation 2006; 114: e498.

²Lubitz SA, Yin X, McManus DD, et al. Stroke as the Initial Manifestation of Atrial Fibrillation: The Framingham Heart Study. Stroke 2017; 48: 490 - 492. doi: 10.1161/STROKEAHA.116.015071.

³Bang CN, Greve AM, Kober L, et al. Incident atrial fibrillation and heart failure in treated hypertensive patients with left ventricular hypertrophy. The LIFE Study. Explor Med 2022; 3: 139 – 148. DOI: https://doi.org/10.37349/emed.2022.00080

⁴Noseworthy PA, Kaufman ES, Chen LY, et al; American Heart Association Council on Clinical Cardiology Electrocardiography and Arrhythmias Committee; Council on Arteriosclerosis, Thrombosis and Vascular Biology; Council on Cardiovascular and Stroke Nursing; and Stroke Council. Subclinical and Device-Detected Atrial Fibrillation: Pondering the Knowledge Gap: A Scientific Statement from the American Heart Association. Circulation 2019; 140: e944 - e963. doi: 10.1161/CIR.0000000000000740.

⁵Shahid S, Iqbal M, Saeed H, et al. Diagnostic Accuracy of Apple Watch Electrocardiogram for Atrial Fibrillation: A Systematic Review and Meta-Analysis. JACC Adv 2025; 4: 101538. doi: 10.1016/j.jacadv.2024.101538.

⁶Cho S, Eom S, Kim D, et al. Artificial intelligence-derived electrocardiographic aging and risk of atrial fibrillation: a multi-national study. Eur Heart J 2025; 46: 839 - 852. doi: 10.1093/eurheartj/ehae790.

⁷Sánchez R, Coca A, de Salazar DIM, Aet al; LASH Guidelines Task Force Steering and Writing Committee. 2024 Latin American Society of Hypertension guidelines on the management of arterial hypertension and related comorbidities in Latin America. J Hypertens 2025; 43: 1 - 34.

⁸American Diabetes Association Professional Practice Committee. Diagnosis and Classification of Diabetes: Standards of Care in Diabetes – 2025. Diabetes Care 2025; 48(Suppl. 1): S27 – S49.

⁹Cuesta G, Pin Arboledas J, Santa María Cano J, et al. Grupo Español de Sueño (GES). Consenso Nacional sobre el síndrome de apneas-hipopneas del sueño. Arch Bronconeumol 2005; 41: 1 - 110.

¹⁰Lang RM, Bierig M, Devereux RB, F, et al ; American Society of Echocardiography's Nomenclature and Standards Committee; Task Force on Chamber Quantification; American College of Cardiology Echocardiography Committee; American Heart Association; European Association of Echocardiography, European Society of Cardiology. Recommendations for chamber quantification. Eur J Echocardiogr 2006; 7: 79 - 108.

¹¹Reddy YNV, Carter RE, Obokata M, et al. A Simple, Evidence-Based Approach to Help Guide Diagnosis of Heart Failure with Preserved Ejection Fraction. Circulation 2018; 138: 861 - 870.

¹²Organización Panamericana de la Salud. Organización Mundial de la Salud. HEARTS en la Américas, Estimar el Riesgo Cardiovascular. Available at https://www.paho.org/cardioapp/web/#/cvrisk Accessed December 26th, 2024.

¹³Segan L, Canovas R, Nanayakkara S, et al. New-onset atrial fibrillation prediction: the HARMS2-AF risk score. Eur Heart J 2023; 44: 3443 – 3452

¹⁴Liao JN, Lim SS, Chen TJ, et al. Modified Taiwan Atrial Fibrillation Score for the Prediction of Incident Atrial Fibrillation. Front Cardiovasc Med 2022; 8: 805399. doi: 10.3389/fcvm.2021.805399.

¹⁵Thompson MQ, Theou O, Tucker GR, et al. FRAIL scale: Predictive validity and diagnostic test accuracy. Australas J Ageing 2020; 39: e529 - e536.

¹⁶Andrade RP, Vitorino PVO, Sousa ALL, et al. A Program to Optimize the Detection of Paroxysmal Atrial Fibrillation: The RITMO Study. Arq Bras Cardiol 2024; 121: e20240235. doi: 10.36660/abc.20240235.

¹⁷Wong YM, Chan PF, Lai KPL, et al. Targeted screening of atrial fibrillation using automated blood pressure measurement device with atrial fibrillation detection function, in patients with type 2 diabetes mellitus in primary care setting. Int J Arrhythmia 2024; 25: 5. Doi https://doi.org/10.1186/s42444-024-00112-x

*Investigators list

Argentina: Mauro Ruise, Mildren Del Sueldo, Mónica Marino, Sofía Brondello, Gonzalo Miranda

Perú: Jorge Paul Juárez Llocla, Luis Miguel Norabuena Rossel

Colombia: Claudia Victoria Anchique Santos, Juan Mauricio Cárdenas Castellanos, Ana Múnera Echeverri

México: Héctor Galván Oseguera, Silvia Palomo Piñón, Enrique Díaz Díaz, Humberto Álvarez López, Luis Ángel Rocha Enciso, Patricia Niuriulú

Ecuador: Yan Carlos Duarte Vera

Chile: Diego Celiz, Henry De Las Salas Pérez

Keywords

Hypertension, Atrial Fibrillation, Home Monitoring, Artificial Intelligence