Feasibility of Using Resting Heart Rate and Step Counts From Patient-Held Sensors During Clinical Assessment of Medical Emergencies (FUSE): Protocol for Prospective Observational Study in European Hospitals

Introduction

Background

The decision about the need of an individual to require hospital admission relies on the assessment of patients’ symptoms, signs, past medical history, medication use, diagnosis, social support at home [1-3], and a judgment about the severity of illness based on estimated risk of deterioration in the subsequent hours and days [4].

Abnormalities of vital signs are quantified by comparison with normal measurements on healthy individuals recorded during periods of physiological stability. Within these individuals, the normal measurements vary and are influenced by age, sex, body composition, medication use, and physical condition [5-7]. There is a relation between the size of change from a physiologically normal range and the number of vital signs affected by abnormality and the frequency of adverse events [8]. Abnormality might be scored with generic tools that can be applied to the majority of patients, such as the National Early Warning Score [9], or tools that are disease-specific, such as the CURB65 [10], Blatchford [11], and DECAF (Dyspnea, Eosinopenia Consolidation, Acidemia, and Atrial Fibrillation) [12]. However, these scores can under- or overestimate risk in individual patients [13].

Knowing the values of an individual patient’s vital signs while well might help clinicians to better understand and interpret their changes or trends [14], their relative degree of deviation from the patient’s own normal and get a better estimate of their personal severity of illness. Similarly, mobility has been suggested to be a vital sign [15] and could be easily quantified by step counts. Normal mobility has been shown to be a protective factor [15]. Therefore, the assessment and comparison of step count during illness might be of value.

The general public are increasingly using devices such as smartwatches and mobile telephones that can potentially measure and record vital signs such as heart rate, heart rhythm, and oxygen saturation. Currently, approximately 90% of the population in the United States [16] and 94% in the United Kingdom [17] are smartphone users. Vital signs are also increasingly being recorded at home using conventional sphygmomanometers and pulse oximeters, and these values, if recorded, might also be available.

The aim of this study is to assess the feasibility of using a patient as their own reference for vital signs by assessing the difference between vital signs on admission to acute care with their own vital sign values recorded by his or her wearable device when they are at home and well (Textbox 1).

Objectives

The primary and secondary study objectives are described in Textbox 1.

Methods

Study Design

A pilot study was performed at a single site in the United Kingdom to develop and test procedures for data collection and reporting during the previous 6 months. Based on the learning from this pilot a prospective, international, multicenter, observational study using the flashmob research design will be conducted [18]. Participating units will collect data from patients presenting with acute medical emergencies to emergency departments, acute medical units, and ambulatory emergency care. There will be no randomization, blinding, and treatment allocation.

Population

Patients who present at the emergency department, acute medical department, and ambulatory emergency care unit in secondary and tertiary hospitals in a sample of hospitals in the Netherlands, United Kingdom, Denmark, and Switzerland with an acute medical complaint. Acute medical complaints are nonsurgical admissions commonly treated by internists such as acute physicians, cardiologists, pulmonologists, gastroenterologists, endocrinologists, geriatricians, and similar (Textbox 2).

Study Procedures

All patients aged 18 years and older who present to acute care will be asked if they wear a smartwatch, activity tracker, or wearable that measures heart rate.

Patients will be asked to give informed consent in line with the guidance from the participating countries. This is likely to include written informed consent in the Netherlands, Denmark, and the United Kingdom. Informed consent will be explicit about accessing data from patient-held wearables (Multimedia Appendix 1). Once consent has been obtained, initial data will be entered into the case report form, and the remaining items of the case report form will be completed on the day after the initial presentation or later (Textbox 3). Study documentation will include a guide for investigators to interrogate commonly used wearable devices (Multimedia Appendix 2). Parameters recorded will vary according to the different devices used but should at least include heart rate and step count.

Withdrawal of Individual Participants

If they wish, participants can leave the study at any time for any reason without any consequences. The investigator can withdraw a participant from the study if they consider it is in the participant’s best interests, such as avoiding delay or compromise of urgent medical care.

Statistical Analysis

Sample Size Calculation

Due to the use of the Flash Mob Research design a formal sample size calculation is not possible. We assume that a number of 40 participating hospitals in Europe screening up to 50 patients each would result in a sample size of at least 2000 patients screened. This sample size should be able to answer questions about the proportion of the number of patients presenting with wearable devices and about changes in key vital signs between assessment in the hospital and a period 24 hours 1 week or 1 month previous.

Data Analysis

Data will be analyzed using SPSS (version 28; IBM Statistics). All variables will be assessed for normality. This study will use basic epidemiological data to determine if patient-owned monitoring devices can be used in an emergency care setting to assess the patient’s current severity of illness and compare it with their usual state of health.

Continuous data will be analyzed using the unpaired t test (if the data are normally distributed) or the Wilcoxon rank sum test (if the data are not normally distributed). Changes between measurements will be described as percentage change from the first measurement taken at home before admission as the assumed baseline and as an absolute change. Categorical data will be analyzed using chi-square test. Basic statistics and descriptives will be used to identify the distribution of patient and device characteristics. Missing data will not be imputed.

Associations between the baseline measurements and the specified clinical end points and the absolute and proportional change of physiological measurements and the specified clinical end points will be assessed using linear regression and logistic regression.

Patient characteristics will be expressed as absolute numbers per (sub)category for the total patient population, the discharged patients, and the admitted patients. The heart rate and step count data will be compared for their respective average or median for the different time points. Both will be tested for significant interaction between time, disposition, and change in either heart rate or step count using a generalized linear model.

For secondary exploratory analyses, the correlation between potential outcomes will be tested using Pearson coefficient correlation analysis (r), where the following interpretation will be used: below 0.10 negligible, between 0.10 and 0.39 weakly correlated, between 0.40 and 0.69 moderately correlated, between 0.70 and 0.89 strongly correlated, and above 0.89 very strongly correlated. The percentile change in heart rate and step count will be calculated in comparison to the baseline, which will be assumed to be at 30 days before a hospital visit.

Primary Outcome Measures

The primary outcome of the study will be feasibility as assessed by the proportion of eligible patients with emergency presentations who have devices as well as recruitment and retention rates.

Secondary Outcome Measures

Secondary outcomes measures include measures of epidemiology and physiology:

We will report trends for heart rate and step count, specifically the difference between the heart rate at home 24 hours, 1 week, and 1 month before clinical assessment from the wearable device compared with the first vital signs measured on admission to acute care as both an absolute and a proportional change.

We will also describe the degree of association between changes to vital signs from the community to assessment in acute care services with admission to a general ward or admission to a critical care area or discharge home within 24 hours from the time of the first hospital assessment.

The population wearing or carrying devices will be described in terms of gender, age, ethnicity, level of education, and digital literacy and contrasted with data from national surveys.

Administrative Aspects, Monitoring, and Publication

Handling and Storage of Data and Documents

Data will be anonymously recorded in a research database (ie, Castor Electronic Data Capture System) in line with local data protection regulations, including but not limited to European General Data Protection Regulations (GDPR).

The collected data will be subsequently stored on a digital research environment from the Erasmus MC, a secure server with 2-factor authentication. There are no patient identifiers within the data. Only selected investigators of this study will have access to the server with the data. Data will be stored for 15 years.

Amendments

Amendments are changes made to the research after a favorable opinion by the accredited MEC has been given. All amendments will be notified to the ethics committee that gives a favorable opinion.

Annual Progress Report

The sponsor or investigator will submit a summary of the progress of the trial to the accredited MEC (Medisch Ethische Toetsings Commissie, METC) once the study is completed or, at the latest, after 1 year. Information will be provided on the date of inclusion of the first participant, the numbers of participants included and the numbers of participants that have completed the trial, serious adverse events or serious adverse reactions, other problems, and amendments.

Temporary Halt and (Premature) End of Study Report

The investigator or sponsor will notify the accredited METC of the end of the study within a period of 8 weeks. The end of the study is defined as the last patient’s last visit.

The sponsor will notify the METC immediately of a temporary halt of the study, including the reason of such an action.

In case the study is ended prematurely, the sponsor will notify the accredited METC within 15 days, including the reasons for the premature termination.

Within 1 year after the end of the study, the investigator or sponsor will submit a final study report with the results of the study, including any publications or abstracts of the study, to the accredited METC.

Public Disclosure and Publication Policy

The results of this study will be published in one or more peer-reviewed articles. Publications will be coordinated by the investigators or sponsors. Authorship of publications will be determined in accordance with the requirements published by the International Committee of Medical Journal Editors and in accordance with the requirements of the respective medical journal.

Benefits and Risks Assessment and Group Relatedness

There are no benefits or risks for participants of this study. The burden of participation in this study is low.

Ethical Considerations

Regulation Statement

The study will be conducted in compliance with the principles of the Declaration of Helsinki (version 10, October 19, 2013) and the principal of Good Clinical Practice. Participants are not subject to procedures or required to follow rules of behavior and, as no additional data are collected the Medical Research Involving Human Subjects Act (also known by its Dutch abbreviation WMO) is not applicable to this study. Each principal investigator is responsible for ensuring that the study is performed in accordance with the protocol and Good Clinical Practice.

The review board of the Erasmus University Medical Center, Erasmus MC in Rotterdam, the Netherlands, has reviewed the research proposal. It confirmed that the rules laid down in the WMO, do not apply to this research proposal Medical Ethics Committee (MEC-2022-0795). The protocol (V1.2 17 February 2025) passed the required approvals in the United Kingdom Integrated Research Application System (IRAS 321129).

Recruitment and Consent

Patients eligible for inclusion are asked if they use a smartwatch or other wearables that measure heart rate by a collaborator of the study. If they do they fulfill the inclusion criteria, and they will then be informed about the study procedures and asked to participate after they have read the patient information form. After receiving this form patients will be asked if they will agree to sign written informed consent.

Governance

Hospitals will be recruited across Europe and coordinated by the study steering board. The study steering board consists of the principal investigators and members of the Safer home consortium. Centers are encouraged to collect data on a small cohort of patients before the main study week (May 10-22, 2024). During the main study week, all participating units are asked to collect data on all patients presenting in the study areas during a 24-hour period.

Participating countries will be responsible for their own submission for approval and coordination of research.

Results

Approvals have been completed in 11 sites in Denmark, 17 sites in the Netherlands, 1 site in Switzerland, and 7 sites in the United Kingdom.

The main FUSE study has gone live on the May 15, 2024, in 34 hospitals in the Netherlands (n=17), the United Kingdom (n=7), Denmark (n=9), and Switzerland (n=1). A total of 1137 patients were screened, of these, 452 patients were included and 209 patients had data on at least 2 collection points, allowing for an analysis of trends.

Preliminary analysis will be completed by December 31, 2024, and the main manuscript will be submitted for publication by the end of March 2025.

Discussion

Anticipated Findings

While sensors that can be carried or worn are in widespread use with a project market size in 2023 of in excess of US $4 billion [20], previous research has found little evidence of their use in emergency care [21]. This study aims to determine if there are opportunities to use the data generated by these consumer-held devices to improve patient care. Trends in vital signs and mobility are used to quantify risks to patients and are likely to become more important as the proportion of patients with chronic conditions (and abnormal vital signs during signs of clinical stability) is increasing. For clinical use we anticipate a number of use cases: Assessing patients whose vital signs have not changed in the week previous presentation in an acute care setting might reassure clinicians that the condition is overall stable and that treatment in the community is likely to be safe. Assessing patients whose vital signs have significantly changed in the days or hours before presentation might increase the confidence of clinicians to admit for observation and treatment in the hospital. This type of decision-making is commonly undertaken during an initial observation phase in emergency departments or acute medical unit with patients being observed for a night before committing to a decision about admission or discharge. We anticipate that this type of decision-making could be undertaken during the initial presentation if data on markers of physiological instability is available already at this point.

Limitations

The distribution of wearables in the population might point toward the exclusion of a significant proportion of the population from this type of support: patients who are less financially well off or not sufficiently digitally or health literate might be disadvantaged in a future health care. This data will hence be collected as part of FUSE. While this study focuses on readily available data and will require in future supplementation by detailed qualitative data.

The study is conducted by a group of secondary care clinicians with expertise in medical emergencies. Many of the patients who will be screened and included will have previously presented in primary care. There is no theoretical reason that the results from this study might not be applicable to those presenting in primary care, but this might require future research.

In preparation of the study, we reviewed sales for the top 20 devices in the Dutch and British markets. For these devices, we have created a data dictionary for investigators that will enable them to assist patients in the location of relevant data. With new devices coming online and changes in consumer habits, the frequency distribution of these devices might however be soon outdated.

The study will explicitly not address the quality of data collected in wearables. We will, however, report on the distribution of devices and approval for medical use where applicable. In the same vein, we anticipate that the duration of usage by individuals will affect the measurements. This is something that we might not be able to evaluate in the current setup.

Given that patients will use a range of devices with different methods for data acquisition and different usage patterns our analysis will focus on the comparison of relative changes in parameters for individuals (intraindividual variation). This relies on the assumption that individuals will have a pattern of usage that is constant over time. Further research will be required, especially if it turns out that the variation between the collected parameters turns out to be large in the days before deterioration and presentation to the hospital.

Finally, future health care needs to be equitable. FUSE will include data that will allow us to gauge some aspects of equity, diversity, and inclusion which are crucial for the development of policy and strategy in this field.

Study Impact

Given the funding shortfall in most health economies consumer-owned monitoring will represent a tempting addition for health care providers in primary and secondary care. The supporting tools such as the wearable device dictionary (Multimedia Appendix 2) might be useful to clinicians even without the data from the study.

Conclusions

This article describes a feasibility study. The results should aid power calculations and the design of more definitive observational and ultimately interventional trials.

From a health policy point of view, there is concern that, at present, device ownership might be more common in those who are younger and well off, thus potentially amplifying problems around equitable access to health care and emphasizing the inverse care law. The FUSE study and subsequent research should provide data that might direct manufacturers toward the design of devices that are suitable for the widest range of users. We hope that FUSE will help policymakers and clinical teams in the design of better and safer services around digitally enabled patients and for those who are not (yet) connected to mHealth systems.