The prevalence and incidence of atrial fibrillation (AF) increase with age and is common among older people [1-4]. A recent study estimated that the global prevalence of AF is 59.7 million [4]. Approximately 70% of individuals with AF are aged between 65 and 85 years [5]. AF has been reported to account for 36% of all ischemic strokes, of which 85% are inadequately anticoagulated [6]. If AF is detected early and managed with appropriate anticoagulation therapy, the stroke risk and subsequent stroke-related disability and mortality can be reduced significantly [7]. Unfortunately, it was estimated that 1% of the general population and 1.4% of people aged ≥65 years were living with undiagnosed AF, as reported in a systematic review that combined data from 30 cross-sectional studies (n=122,571) [8]. Hence, the opportunity for anticoagulation therapy to reduce stroke risk for these patients is missed.

Several guidelines have recommended opportunistic screening for AF [9-12]. However, studies suggest that one-off opportunistic screening approaches have a low yield for identifying AF. For example, a recent cluster randomized trial of opportunistic screening using pulse palpation, electronic blood pressure measurement with an AF algorithm, and a handheld single-lead electrocardiogram (ECG) device versus usual care for detection of AF in primary care patients (involving 9218 patients in the intention-to-screen group, 55% women, mean age 75.2 years vs 9526 patients in the usual care group, 54.3% women, mean age 75.0 years) found that opportunistic screening did not improve AF detection compared with usual care [13]. On the contrary, repeated heart rhythm monitoring over a duration increased the yield of AF detection. Petryszyn et al [14] reported in a systematic review that repeated heart rhythm monitoring with ECG devices over periods ranging from 2 weeks to 12 months had higher AF detection rates compared with one-off opportunistic screening approaches: 2.1% (95% CI 1.5-2.8) with repeated ECG screening versus 1.2% (95% CI 0.8-1.6) with opportunistic screening (P<.05). Although many guidelines advocate the use of 12-lead ECG for opportunistic screening, these may limit locations where screening may occur and demand a higher skill level to operate 12-lead ECG devices [15,16]. Mobile single-lead handheld ECG devices are easier to use with better time efficiency compared with 12-lead ECG machines, and these single-lead handheld ECG devices have been used in several AF screening studies [15].

Recent systematic reviews, including 8180 single-lead ECG tracings, show that mobile handheld single-lead ECG devices have high accuracy for diagnosing AF [17]. These devices are available to the public and clinicians. However, a national survey reported that although general practitioners (GPs) are aware of the devices, they rarely conduct AF screening in their busy clinical practice [18]. Clinician-led AF screening faces barriers because clinicians are facing competing clinical priorities and time constraints [19]. Alternative strategies for early detection and management of AF are needed. Patient-led AF screening through self-recording of single-lead ECG using mobile handheld ECG devices could be an alternative. A randomized controlled trial (RCT) involving 7173 community-dwelling older people aged 75 to 76 years in Sweden reported that screening using patient-activated intermittent ECG recordings with a handheld ECG device (Zenicor) twice daily over 2 weeks, when the participants noticed palpitations, increased new AF detection fourfold [20]. Similarly, in an RCT of AF screening using a handheld ECG device (AliveCor Kardia) in 1001 participants aged ≥65 years and with CHADS-VASc (congestive heart failure, hypertension, age ≥75 [double score], diabetes, stroke [double score], vascular disease, age 65 to 74 and sex category) score ≥2, the detection rate of new AF after 12 months was 3.8% in the monitored group versus 0.1% in the control group (hazard ratio 3.9, 95% CI 1.4-10.4; P=.007) [21]. Another RCT of AF screening using a 2-week ambulatory ECG patch (Zio XT), one at baseline and another at 3 months in 856 participants aged ≥75 years with hypertension, increased new AF detection rate 10 folds (5.3% in the monitored group vs 0.5% in the control group; relative risk: 11.2; 95% CI, 2.7-47.1; P=.001) [22]. These studies [20-22] suggest the potential of patient-led approaches to AF screening but still leave questions on how to implement these approaches and the generalizability of these approaches. Unanswered questions about the implementation of AF screening programs include whether such programs can be implemented alongside existing health care systems; whether regular self-screening with mobile devices is feasible and acceptable in older adults; what is the feasibility, resource use, and clinician acceptability of real-world implementation of such programs; what is the time taken by services overseeing and monitoring such programs in terms of reviewing and interpreting large amounts of ECG data; and what strategies can be applied to optimize the use of resources. In addition, there is less data on the barriers to and enablers of the implementation of such programs, longer self-monitoring periods, and implementation in subgroups (such as older people who are frail and people living in remote areas) in which these strategies may not work. A recent systematic review reported that the prevalence of frailty in patients with AF was up to 75% [23]. More studies are needed to better assess whether such mobile health devices can be used effectively and implemented in programs at a large scale among older people who are frail.

There is also a lack of information about the role and importance of patient empowerment with respect to the implementation of AF screening. The World Health Organization promotes patient empowerment (ie, training patients to perform and engage in health-related behaviors within their familiar setting) as it can potentially lead to positive health outcomes [24]. Patient empowerment can be incorporated in patient-led AF screening by training patients to self-record single-lead ECGs. However, patient empowerment has its limitations; that is, patients face automated ECG interpretation results that are often beyond their competence to understand and act upon, and it is impractical to have every ECG result individually and regularly checked by their clinicians. A centralized monitoring system is a feasible way of remotely monitoring a patient’s heart rhythm [25].

The processes of the screening program can vary in their actual implementation because of diverse contexts and participant characteristics (both patients and clinicians); for example, participants may be incapable of or not engage in performing self-recording of ECG, or they may not follow up with (or do not have access to) their clinician after a clinically significant abnormality is detected and notified. Process variations can affect outcomes. Therefore, it is important to evaluate the processes with the aim of better understanding why variations occur and how to improve the processes to achieve an effective intervention and identify contextually relevant strategies to scale up the screening program to benefit larger populations.

In summary, there are gaps in our knowledge regarding the feasibility, effectiveness, and acceptability of patient-led AF screening by remote patient self-recording with centralized clinician-supported monitoring of single-lead ECGs in older community-dwelling people who are frail. The Mass AF screening program (Figure 1) is designed for implementation among community-dwelling people aged ≥75 years. It comprises the provision of a handheld ECG device and training of participants to self-screen on weekdays and transmit ECGs for review by a central monitoring team. We aim to implement and evaluate this AF self-screening program in which older people in the community are empowered to perform repeated heart rhythm monitoring using a single-lead handheld ECG device and connected with health care providers who review and support the diagnosis of AF and management by primary care and specialist services. We hypothesize that the proposed self-screening model of care may lead to several positive outcomes, including a feasible and scalable model for implementing patient-led AF screening in community-dwelling older people, improved patient satisfaction by empowering them with the relevant knowledge and skills to perform self-screening [24], and thereby higher adherence to the screening program [26,27].

Our study objectives are to (1) compare AF ascertainment rates in the intervention and control groups; (2) evaluate the feasibility of the intervention, including assessing participant satisfaction, acceptability, barriers, and enablers and how frailty influences these assessments; and (3) assess agreements between the ECG device automatic algorithm and clinician interpretation.

Alongside these objectives, the specific objectives of the process evaluations are as follows:

This is a 2-arm, randomized, open-label, waitlist-controlled trial in community-dwelling people aged ≥75 years. We will also conduct a process evaluation of this study. Randomization in the ratio of 1:1 is stratified by participant frailty status (frail or nonfrail; Multimedia Appendix 1). Participant frailty was determined using the FRAIL (Fatigue, Resistance, Ambulation, Illnesses, and Loss of Weight) scale based on five components: fatigue, resistance (inability to climb stairs), ambulation (inability to walk a certain distance), illness, and loss of weight [28,29]. The intervention group will commence the monitoring program for 12 months upon enrollment. The control group will be waitlisted for the first 6 months and then commence the monitoring program in the subsequent 6 months. The steps involved for enrollment, randomization, intervention, control, and exit from the program are outlined in the study flowchart (Figure 2), and descriptions of the screening program are provided in the following sections.

Our target population involves older people living independently in the community outside of a hospital, nursing home, or similar institutional residence. The inclusion criteria are as follows: community-dwelling people aged ≥75 years, having a smartphone or electronic device that can operate the AliveCor Kardia mobile app, and being able to understand instructions in English. Individuals with the following conditions will be excluded: previously confirmed diagnosis of AF, having an implantable cardiac monitor, pacemaker or defibrillator, dementia, inability to provide informed consent, and those with a medical illness with an anticipated life expectancy of <3 months.

Participants allocated to the intervention will immediately commence the monitoring program. They will be provided with a small handheld single-lead Kardia ECG device (AliveCor Inc), which has been cleared by the Food and Drug Administration and approved by the Therapeutics Goods Administration. After the participants receive the device, they will receive a phone call from research assistants who will help them to set up the device, including downloading the Kardia app to the mobile phone, setting up reminders to record ECG in the app, and setting up a Kardia user account. The research assistants will create a participant profile in the central monitoring portal, which will generate a unique 12-digit code for the participant. Research assistants will inform participants of their unique 12-digit code via SMS text message or over the phone. The participant will enter the 12-digit code in their Kardia user account, and once this step is completed, the Kardia user account will be connected to the central monitoring portal. Participants can commence recording the ECGs, which will be transmitted to their personal profiles in the central monitoring portal.

Research assistants will take participants through the steps of recording an ECG. Participants will record a single-lead ECG trace by placing 2 fingers of each hand steadily on 2 small touchpads (3 cm × 3 cm) of the device for 30 seconds on weekdays. The ECG device will be connected to the mobile phone wirelessly via the Kardia app that they have downloaded. An ECG trace will appear on the participants’ mobile phones, and participants will be able to record notes in the ECG trace. We will encourage them to note the activities they performed before recording the ECG. The ECG and notes will be automatically transmitted to the central monitoring portal. The training conducted over the telephone or video call between research assistants and participants will take approximately 30 to 60 minutes per participant. In the context of the COVID-19 pandemic, all study-related procedures will be conducted remotely using phone calls or video calls (if participants have access to and prefer this modality). To ensure that participants are confident in using the device, research assistants will call each participant to confirm that they are able to record an ECG. Participants will be encouraged to have a family member who may assist them in the process. After the training, participants will also receive an SMS text message with a weblink to an instructional video created by the device manufacturer. The short training video will serve as a reference for participants to refresh their memory on how to record a single-lead ECG. The research team will contact participants if they have not recorded any ECGs for 3 consecutive days to find out and address the causes, if possible.

Before commencing the study, participants will be advised that in the event of experiencing symptoms (eg, syncope, chest pain, palpitations, and shortness of breath) that are severe in nature or that are of concern to the participants, to present to their local medical physician or hospital for assessment as soon as possible.

Participants allocated to the control group will have usual care and be told that they have been waitlisted to start the monitoring program in 6 months. During the 6-month waiting period, it is expected that participants in the waitlist group will visit their GPs as per their usual health care needs, and their GPs will provide care and referrals as usual.

In computing the sample size required to assess the primary feasibility outcome, we will evaluate the proportion of participants reporting being satisfied or very satisfied that their heart rhythm was monitored in the past 6 months in the intervention group versus the control group. We arbitrarily set that 50% of the participants in the control group will be satisfied or very satisfied. With reference to the literature that reported a proportion of 67% [30] to 82% [31] of older people were satisfied or very satisfied with the use of technology-enabled monitoring at home, we postulate that there will be an absolute 30% increase in satisfaction in the intervention group compared with the control group. Our study will have 80% power, using a 5% level of significance, to detect an absolute difference of 30% in satisfaction between the 2 groups. A sample size of 100 participants aged ≥75 years is required to assess the primary feasibility outcome

To calculate the sample size required to evaluate the primary clinical outcome of AF detection rate, we set an AF detection rate of 10% in the intervention group and 1% in the control group, in accordance with a recent study [22]. At 80% power, a 2-sided test, and α .05, we estimate that a sample of 200 participants will be needed to detect a significant difference in AF detection between the intervention and control groups. Therefore, a total of 200 participants will be recruited for this trial to assess the primary clinical outcome.

Participants will be randomized to the intervention or control group on a 1:1 basis stratified by baseline measure of frailty (ie, frail or not frail according to the FRAIL score) [28,29] and using permuted blocks of sizes 4 and 6. The statistician has generated a randomization list using the RandomiseR package in R software (R Foundation for Statistical Computing) [32]. The randomization list will be input into the REDCap (Research Electronic Data Capture) [33] database, which captures participant demographic and baseline data and survey findings (Multimedia Appendix 1). The statistician and principal investigators will be unaware of patient allocation until after the completion of the study.

A multipronged approach will be used to identify potential participants. We will use clinician networks (eg, GPs, cardiologists, geriatricians, and allied health professionals) and a variety of direct approaches to the community to recruit a wide spectrum of participants from various demographic backgrounds living in wide geographical areas across Australia. Communications will be sent to practice managers seeking their assistance in disseminating introductory letters, leaflets, flyers, and posters to their clinicians. The practice managers may also disseminate the information to their patients through their usual channels of communication, including displaying them in waiting rooms, websites, newsletters, or in electronic format or hard copy. The decision to contact the research team will be at the discretion of the patients. In addition, a direct community approach will be used. The research poster and leaflets will be disseminated in local community centers such as the Returned and Services League Australia and places of worship. People interested in the study will initiate contact with the study team directly by email or phone to receive further information. We will also list the study with third-party recruiters such as HealthMatch [34] and Join US [35]. Individuals who contact the research team will be screened for eligibility and provided with further explanations about the study. We will inform GPs about their patients who enroll in the study.

The research team will confirm the eligibility of interested individuals against the inclusion and exclusion criteria over the phone. Eligible individuals will be provided with participant information statements and consent information and given time to read the information. Research personnel will answer individuals’ questions. Participants will provide verbal consent to a member of the research team who will electronically sign off the consent form and keep the form in the secured university computer drive.

Qualified and trained study personnel, including a cardiac technician and clinical monitoring personnel with medical qualifications, will remotely review all ECGs and compare their diagnosis with the device’s automated diagnostic algorithm. If the ECGs are normal or have minimal abnormalities and the personnel are certain of their diagnosis, the ECGs will not be referred to a cardiologist. However, the personnel will refer all abnormal ECGs for diagnosis confirmation, or uncertain ECG abnormalities for clarification of diagnosis, to cardiologists or cardiac electrophysiology specialists. Participants’ ECGs will be classified into low, moderate, high, and severe abnormalities and managed as shown in Table 1. The research team will notify the participants and their GPs of AF or other clinically significant ECG abnormalities. A copy of an abnormal ECG will be forwarded to their GP. When an AF diagnosis is confirmed by the research team, the participant will be advised to see their GP. The participant will exit the screening program or opt to continue the monitoring program. The research team will contact the GP to obtain information about the treatment given to the patient.

All study procedures have been designed to be conducted remotely using telephone or video calls. At baseline, we will obtain information on sociodemographics, self-reported weight and height, and concurrent medical conditions and medications and data to assess stroke risk, frailty, and activities of daily living. At the end of the program, we will conduct a usability survey of all participants via phone calls to obtain information related to their experiences with the screening program and obtain further information on any adverse events while participating in the program (Multimedia Appendix 1). All GPs who have patients enrolled in the study will be invited to provide their feedback in a survey (Multimedia Appendix 2). All electronic data and documents related to participants and the project will be securely stored in the university computer drive accessible by authorized research team members only.

We have followed the Medical Research Council guidelines in designing the process evaluation [36]. Participants’ and GPs’ expectations and experiences of the screening program may be influenced by various contextual factors such as the participants’ social and cultural background and the GPs’ clinical practice resource and setting. Using a theoretical lens of critical realism [37], we will provide an explanatory analysis of the perceptions, experiences, and interactions with contextual factors that participants describe (ie, what works for whom and under what circumstances).

After participants have used the handheld ECG device for a minimum of 3 months, the research team will invite several participants in both groups (≥10 participants to achieve thematic saturation) to attend an in-depth semistructured interview to explore their views and feedback on the study. The semistructured interview will explore the participants’ feedback on their experiences with using the ECG device, use of the device for detecting irregular heart rhythm, participants’ perceptions of this remote screening method, and their access to health services generally (Multimedia Appendix 1). The invitation will be based on purposive sampling to obtain representative participants from a wide spectrum of demographic groups, that is, male or female, rural or urban, and frail or not frail.

GPs will be invited to a one-on-one in-depth interview, which will take approximately 20 to 30 minutes. The interviews will be audio recorded and transcribed by the research team. We plan to recruit at least eight GPs for this interview. A sample of 8 GPs is considered appropriate for the exploratory interviews [38]. However, more GPs will be recruited if thematic saturation is not achieved. The semistructured interviews will explore GPs’ views with respect to AF screening generally and in screening people aged ≥75 years particularly, their views of the patient-led self-screening in this research study, their knowledge of the use of mobile health devices (including handheld ECG devices) in clinical practice, and their views on using the handheld ECG device (ie, AliveCor Kardia) for AF screening in this study. In alignment with a critical realism [37] approach, the interview questions will explore GPs’ contextual factors and their interaction with patient participants and the research team with the aim of exploring GPs’ perception of their roles and barriers to and acceptability and enablers of this patient-led self-recording of a single-lead ECG program (Multimedia Appendix 2).

The clinical outcomes include (1) new AF detected over 6 and 12 months and (2) appropriate use of anticoagulant therapy. The feasibility outcomes include (1) participant satisfaction that their heart rhythm was monitored in the past 6 months; (2) participant recruitment rate; (3) frequency of ECG transmission to the central monitor; (4) proportion of participants who complete the program; (5) proportion of dropouts (exit the program prematurely) and reasons; (6) actual costs of the screening program; (6) agreement between the ECG device’s automatic algorithm and clinician interpretation; (7) usability assessment; (8) participant acceptability and barriers to and enablers of implementation; and (9) impact of frailty on feasibility assessments and outcomes.

All analyses will be conducted according to the principle of intention-to-treat. Continuous variables will be presented as mean (SD) or median with IQR and categorical variables as frequency and percentage. The key outcomes, namely participant satisfaction in the AF screening program, usability of the screening program, and the incidence of new AF detected over 6 and 12 months, will be reported as frequency and percentages and will be compared between the intervention and control groups using the chi-square test. Comparisons between groups (intervention and control) for other outcomes will be assessed using the chi-square test or Fisher exact test for categorical variables and Student t test or Mann-Whitney test for continuous variables as appropriate. We will consider 2-tailed P<.05 as statistically significant. The agreement between the ECG device automatic algorithm and clinician interpretation will be evaluated using κ statistics. Subgroup analyses will be performed based on age, gender, and location. Subgroup analysis by frailty will be performed to examine the potential impact of frailty on the outcomes, feasibility, and acceptability of the program. The actual operational costs of delivering the project will be recorded and verified using invoices and receipts, including the costs of personnel involved in interpreting ECGs, which will be computed based on their hourly wages and the time they spent in their roles in this program. The resultant costs will be compared with the costs reported in the literature on the costs associated with the detection of AF in a similar older population. Missing data will be identified, and its causes will be described. Sensitivity analysis will be performed to examine the robustness of the findings [39].

Qualitative evaluation will be reported according to the COREQ (Consolidated Criteria for Reporting Qualitative Research) guidelines [40]. Interviews with patients and GPs will be thematically analyzed using an inductive approach [41]. Themes will be interpreted through the critical realism lens [37] and compared with the literature [42,43]. We will triangulate the quantitative and qualitative findings [44] from patient participants and GPs to acquire an in-depth understanding of the barriers to and enablers of implementing the screening program.

The process evaluation components [36], explanatory data, and anticipated outcomes are summarized in Table 2.

The quantitative and qualitative analysis approach and methods are summarized in Textbox 1.

This study was approved by the human research ethics committee of the University of Sydney (reference number 2020/680). The study is conducted in full conformance with principles of the International Committee on Harmonization of Good Clinical Practice and Declaration of Helsinki Good Clinical Practice guidelines and within the laws and regulations of the Australian National Health and Medical Research Council.

This patient-led AF screening in the community is different from clinician-led opportunistic screening. In this model of screening, participants are trained and empowered to self-record ECGs instead of awaiting clinicians to screen them opportunistically. The centralized remote monitoring team will facilitate patient access to see their GPs.

Drawing on the strengths of quantitative and qualitative methodologies [45], this study will provide evidence for AF detection rates, participant satisfaction, and feasibility of implementing this program using a telephone, a video interface, and the internet for older people, including people who are frail, with the potential to extend to other vulnerable groups such as people with disabilities, people who are socially isolated or because of the COVID-19 pandemic lockdown, and those who live in remote areas.

Participant satisfaction scores often reflect the convergence and gap between participant expectations and actual experiences [46], and satisfaction scores also measure how well the intervention was received by participants [31]; hence, satisfaction is a commonly evaluated outcome in clinical trials [46]. Nonetheless, satisfaction scores would not provide insights into participant experiences, which provide contextualized feedback to improve the screening program. We complement this with in-depth one-on-one semistructured interviews with participants and GPs to explore their insights on barriers and enablers.

This is a prospective RCT design. The waitlist-controlled design provides equitable access to all participants in a mass screening strategy. It is a simple, acceptable, and noninvasive screening strategy that can be implemented regardless of geographical location. Our novel approach in promoting patient-empowered self-screening integrated with a clinician-coordinated centralized system will provide patients with integrated care that facilitates access to GPs and specialist services. Patients will receive training to use the device, and they will be reminded to perform their routine ECGs if they have not done so for 3 consecutive days. This type of interaction and reminder system has been proven to yield positive health outcomes such as engagement in positive health behaviors [47]. Performing self-screening could raise awareness of self-care and improve patient confidence in self-care, which is a form of patient empowerment that is promoted by the World Health Organization [24].

We anticipate that this study will provide data on whether implementation of this type of community-based model of care is feasible and acceptable to patients and health providers in the community. At study completion, the results will be shared with the Heart Foundation (study funder), policy makers, health providers, consumers, and other stakeholders. Access to ECG monitoring devices for future screening programs is dependent on feasibility from a cost perspective, as well as aspects of whether this would be a barrier to implementation. We will conduct a qualitative analysis to understand participants’ perceptions of the value of the monitoring device to their well-being, as well as the affordability of the device.

This study will provide information on the usability of and costs associated with AF mass screening in Australian people aged ≥75 years. It will also provide evidence of AF incidence in older people in Australia. This can potentially facilitate the development of a national screening program for AF in older people and people who are frail.

Screening for AF is more likely to occur in the community or general practice setting than in the hospital setting. It took an average of 10.6 minutes to acquire a 12-lead ECG in a general practice setting (including the time preparing the patient for ECG acquisition and placing the electrodes correctly on the patient) [48]. In contrast, this patient-empowered self-screening potentially reduces time constraints faced by clinicians as patients are empowered to self-record ECGs in the community rather than clinicians spending the additional time acquiring ECGs opportunistically in a busy clinical setting. In this program, GPs can access help from the participating cardiology team to confirm the diagnosis and facilitate appropriate management of new AF if necessary. This could enhance access and interaction between GPs and cardiologists in providing integrated care to patients to achieve better health outcomes.

All study procedures have been designed to be conducted remotely by telephone or video interface to facilitate this study in the context of the COVID-19 pandemic. Therefore, the collection of baseline data and medical history relies on participants’ self-reported information. There is a potential for loss to follow-up in older patients who are frail; for example, participants do not record ECGs. We will attempt to minimize this by following up with patients when no ECG is received for 3 consecutive days. There may be a potential selection bias in this study based on the inclusion criteria. For example, this trial is limited to older people who can understand English and have a smartphone. These participants may come from higher socioeconomic communities and represent a more motivated cohort than the general population of older people.

The findings from this implementation study will guide the development of practical and attainable solutions to address a gap in AF screening among older people in the community and other vulnerable groups. In addition, this study will explore the experiences and feedback from participants and clinicians and provide new knowledge on the processes involved in the implementation of the screening program and how processes can be improved, replicated, and scaled up to reach larger populations.