Accuracy of tests for detection

National Guideline Centre (UK)

Accuracy of tests for detection

Atrial fibrillation: diagnosis and management

Evidence review B

NICE Guideline, No. 196

Authors

National Guideline Centre (UK).

London: National Institute for Health and Care Excellence (NICE); 2021 Apr.

ISBN-13: 978-1-4731-4043-1

1. Detection diagnostic accuracy

1.1. Review question: What are the most accurate methods for detecting atrial fibrillation in people with cardiovascular risk factors for AF and/or symptoms suggestive of AF?

1.2. Introduction

Please see Evidence review A.

1.3. PICO table

For full details see the review protocol in Appendix A:.

Table 1

PICO characteristics of review question.

1.4. Methods and process

This evidence review was developed using the methods and process described in Developing NICE guidelines: the manual.¹⁷⁴ Methods specific to this review question are described in the review protocol in Appendix A:.

1.5. Clinical evidence

1.5.1. Included studies

Seventy four studies were included in this review.⁶^, ⁷^, ²³^, ²⁴^, ²⁶^, ³⁶^, ⁴⁹^, ⁵⁸^, ⁵⁹^, ⁶³^, ⁷⁶^, ⁷⁷^, ⁷⁹^, ⁸²^, ⁸⁶^, ⁹⁰^, ⁹¹^, ⁹⁶^, ¹⁰¹^, ¹⁰⁴^, ¹¹⁷^, ¹²³^, ¹²⁶^, ¹²⁸^, ¹³²^, ¹³³^, ¹³⁸^, ¹⁴⁰^, ¹⁴⁴^, ¹⁴⁵^, ¹⁵⁰^, ¹⁵³^, ¹⁵⁶^, ¹⁶⁰^–¹⁶²^, ¹⁶⁴^, ¹⁶⁵^, ¹⁷¹^, ¹⁷²^, ¹⁷⁷^, ¹⁸⁴^, ¹⁸⁶^, ¹⁹⁵^–¹⁹⁷^, ²⁰¹^, ²⁰⁸^–²¹⁰^, ²¹⁴^, ²¹⁸^–²²⁰^, ²²²^, ²³³^, ²³⁷^, ²⁴⁰^, ²⁴³^, ²⁵³^, ²⁵⁸^, ²⁶⁵^, ²⁶⁸^, ²⁷¹^, ²⁷⁵^, ²⁷⁸^–²⁸¹^, ²⁸³^, ²⁸⁴^, ²⁸⁶^, ²⁸⁸^, ²⁹⁵

The characteristics of these studies are summarised in Table 2 and Table 3, and evidence from these studies are summarised in the clinical evidence summaries (Table 4 to Table 14). Further details are available in the study selection flow chart in Appendix C:, sensitivity and specificity forest plots and receiver operating characteristics (ROC) curves in., and study evidence tables in Appendix D:.

Analysis was stratified by the gold standard used in the studies: 1)12 lead ECG interpreted by an expert (such as a cardiologist or electrophysiologist) or 2) ambulatory monitoring for >24 hours (such as Holter). This stratification was based on the AF that would be detected. 12 lead ECG should detect persistent AF but will only pick up paroxysmal AF during specific intervals of time, and is therefore only a valid gold standard for persistent AF. Ambulatory monitoring for >24 hours may be more useful at picking up AF of both persistent and paroxysmal types and so can be used as a more valid gold standard for any type of AF. Table 2 provides details of the reference standards used.

For each of the above separate strata, pre-hoc sub-grouping strategies (conditional on observed heterogeneity) for any diagnostic test meta-analyses were by

1): expertise of index test interpreter (automated reading / expert reader [such as cardiologist or electrophysiologist] / clinician [clinician such as nurse or GP that was not deemed to be an expert in analysis of ECG traces] / patient).

2): simultaneity of index and reference tests (yes/no)

Sub-grouping was only carried out for the ‘Alive Cor’ test because this was the only analysis where heterogeneity was evident and where there would be at least 3 studies in a sub-group. For the ‘AliveCor’ test, sub-grouping was carried out using the ‘expertise’ strategy and not the ‘simultaneity’ variable because there was evidence from the data that only the former sub-grouping variable could explain the original heterogeneity.

Only 6 diagnostic meta-analyses were possible because at least 3 studies are required for a valid pooling of results, and for most index tests only one or two studies were available. Where diagnostic meta-analysis was possible for a particular test, data from the same study that involved different interpreters were considered as separate data points. Such data were therefore entered alongside each other in the meta-analysis. This was necessary because expertise of examiners had been classified as a ‘sub-grouping’ (conditional stratification) variable rather than a ‘stratification’ (unconditional stratification) variable in the protocol. This meant that we could only stratify the meta-analysis by the expertise of interpreters if there was observable heterogeneity in the initial non-stratified analysis. This inclusion of more than one data point from the same study in the meta-analysis was not deemed to be ‘double-counting’ for two reasons. Firstly, the use of interpreters of different expertise was felt to make data points from the same study sufficiently ‘different’ to each other to the extent that they could be regarded as being from ‘different studies’ for the purposes of meta-analysis. Secondly, in many cases the samples of patients used for different interpreters within the same study were different or only overlapped partially.

In the vast majority of studies the unit of analysis was the person being tested, and if AF was detected once in that person then this was counted as a positive test result (regardless of how many times AF was detected in that person using that test) in the 2x2 table. This reflects the purpose of the tests – to find out if a specific patient has AF or not, and as soon as AF has been detected a diagnosis may be made. However in 5 studies¹⁵³^, ¹⁷¹^, ²¹⁸^, ²⁶⁸^, ²⁷⁵, the unit of analysis was each of many separate measures done on each person (person-measures). Thus, if AF was detected on several occasions on one person, each event was considered a separate positive test. Since this may influence the strength of overall results, care should be taken with interpretation of these results. Therefore, where such results occur this has been highlighted (sections 1.5.6 and 1.5.7).

Most studies did not include the exact protocol population. For example, some studies contained people without symptoms suggestive of AF. Such studies were included with a quality downgrade for ‘indirectness’, as stated in the protocol. This flexibility was useful because very few studies were available that exactly met the protocol’s population requirements. Furthermore, it was felt that the sensitivity and specificity of the devices would not be greatly influenced by variations in population characteristics, as it was felt implausible that any of these varying characteristics could significantly affect how easy it is to detect AF. It was accepted that different populations would have different prevalence of AF, and that this would therefore affect positive and negative predictive values. However, rather than to directly evaluate predictive values, the clinical aim of this review was to assess the sensitivity and specificity of tests, which independently measure their clinically important ability to differentiate people who have and who don’t have the condition. Nevertheless, it was recognised that positive and negative predictive values are of great importance to health economic analysis, and so these will be calculated from the sensitivity and specificity data from the studies in conjunction with established UK prevalence rates (rather than the prevalence rates in individual studies) if tools are found with strong evidence of adequate sensitivity and specificity. Similarly, although ‘screening’ is outside the remit of this review, diagnostic papers with a reference to screening were included if they contained useful data on the accuracy of tests. The rationale for this is that the determined accuracy of a single device would be similar, whether it is part of a screening strategy or not.

Finally, there were some features of some of the data that should be clarified.

Occasionally, papers reported some data from the index test as unclear, and varied in whether they designated this as ‘AF’ or ‘non-AF’. For the purposes of this review, any such data were designated ‘non AF’, regardless of how the paper designated the data. This approach was taken because this review is about detection of AF. If a data point is unclear then AF cannot be said to have been detected, so in a binary classification system it can only be designated ‘non-AF’. However, if unclear data in papers were only designated as AF, and there was insufficient information in the paper to allow re-calculation, those data were used.
Sometimes a paper might have several index test interpreters who were at the same level of expertise (for example cardiologist 1, cardiologist 2, etc.) but their data were considered separately. In such cases only the first reported observer was included in this review, to avoid ‘double counting’ of similar data.
Destegne, 2017⁵⁸ provided data for a sample including people with pacemakers or implanted cardiac monitors, as well as data for a sample with such people excluded. The latter sample was used for this review as people with pacemakers or implanted cardiac monitors were not part of the population in other studies, and had a significant effect on results

1.5.2. Excluded studies

Please see the excluded studies list in Appendix H:.

1.5.3. Summary of clinical studies included in the evidence review (Gold standard = 12 lead ECG stratum)

Table 2

Summary of studies included in the evidence review for detection of atrial fibrillation.

1.5.4. 1.5.5. Summary of clinical studies included in the evidence review (Gold standard = >24 hours ambulatory monitoring stratum)

Table 3

Summary of studies included in the evidence review for detection of atrial fibrillation.

See Appendix D: for full evidence tables.

1.5.6. Quality assessment of clinical studies included in the evidence review

For measurement of imprecision, clinical decision thresholds for sensitivity and specificity were set at 0.90 and 0.60.

STRATUM 1: 12 lead ECG interpreted by expert cardiologist/electrophysiologist as gold standard

Table 4

Clinical evidence summary: diagnostic test accuracy for mobile ECG devices (12 lead ECG interpreted by expert cardiologist/electrophysiologist as gold standard). Where 95% CIs are provided in round brackets (or no 95% CIs are given), raw data were not (more...)

Table 5

Clinical evidence summary: diagnostic test accuracy for blood pressure monitors (12 lead ECG interpreted by expert cardiologist/electrophysiologist as gold standard). Where 95% CIs are provided in round brackets (or no 95% CIs are given), raw data were (more...)

Table 6

Clinical evidence summary: diagnostic test accuracy for pulse palpation (12 lead ECG interpreted by expert cardiologist/electrophysiologist as gold standard). Where 95% CIs are provided in round brackets (or no 95% CIs are given), raw data were not available (more...)

Table 7

Clinical evidence summary: diagnostic test accuracy for photoplethysmography (12 lead ECG interpreted by expert cardiologist/electrophysiologist as gold standard).

Table 8

Clinical evidence summary: diagnostic test accuracy for 3-lead tele ECG (12 lead ECG interpreted by expert cardiologist/electrophysiologist as gold standard). Where 95% CIs are provided in round brackets (or no 95% CIs are given), raw data were not available (more...)

Table 9

Clinical evidence summary: diagnostic test accuracy for 6 lead ECG (12 lead ECG interpreted by expert cardiologist/electrophysiologist as gold standard). Where 95% CIs are provided in round brackets (or no 95% CIs are given), raw data were not available (more...)

Table 10

Clinical evidence summary: diagnostic test accuracy for other non-12 lead ECG (12 lead ECG interpreted by expert cardiologist/electrophysiologist as gold standard). Where 95% CIs are provided in round brackets (or no 95% CIs are given), raw data were (more...)

Table 11

Clinical evidence summary: diagnostic test accuracy for 12 lead ECG interpreted by automated algorithm or non-expert interpreters (12 lead ECG interpreted by expert cardiologist/electrophysiologist as gold standard). Where 95% CIs are provided in round (more...)

1.5.7. Quality assessment of clinical studies included in the evidence review

STRATUM 2: >24 hour ambulatory monitoring [such as Holter] as gold standard

Table 12

Clinical evidence summary: diagnostic test accuracy for blood pressure monitors (>24 hour ambulatory monitoring as gold standard). Where 95% CIs are provided in round brackets (or no 95% CIs are given), raw data were not available and Forest Plots (more...)

Table 13

Clinical evidence summary: diagnostic test accuracy for <7 day Holter devices (7 day Holter as gold standard). Where 95% CIs are provided in round brackets (or no 95% CIs are given), raw data were not available and Forest Plots or pooled analyses (more...)

Table 14

Clinical evidence summary: other longer term devices (>24 hour ambulatory monitoring as gold standard). Where 95% CIs are provided in round brackets (or no 95% CIs are given), raw data were not available and Forest Plots or pooled analyses were (more...)

1.6. Economic evidence

Please see evidence review A.

1.7. The committee’s discussion of the evidence

1.7.1. Interpreting the evidence

1.7.1.1. The outcomes that matter most

For the diagnostic accuracy review, the outcomes were sensitivity and specificity. For a test that is suitable to be used alone as a definitive diagnostic test (in place of 12 lead ECG), both sensitivity and specificity are of equal value, as a definitive test needs to have almost perfect sensitivity and specificity. High sensitivity is essential to avoid people with true AF being missed and therefore untreated, as this can lead to serious sequelae such as stroke. High specificity is equally important to prevent people without AF being misdiagnosed as having it, which may lead to unnecessary prescription of anticoagulants, antiarrhythmic drugs, or invasive procedures, all of which carry a burden of serious adverse effects.

In contrast, for tests that might be used as the first part of a two stage testing process (an example of such a two stage process is pulse palpation followed by 12 lead ECG in people who test positive) then sensitivity may be more important than specificity. Reasons for this are as follows. In a two-test scenario, the initial test is used as a filter to decide who goes on to the resource-intensive 12 lead ECG, and this could be achieved by either an extremely sensitive initial test or an extremely specific initial test. With a highly sensitive initial test, only initial positives go on to the next stage of testing (where the false positives resulting from the sub-optimal specificity of the initial test can be ‘weeded out’ by 12 lead ECG). Initial negatives can be safely discarded from the diagnostic process when the initial test has high sensitivity, because very high sensitivity means that the initial negatives should contain hardly any people with true AF. In contrast, with an extremely specific initial test, only initial negatives go on to further testing (where the false negatives resulting from the sub-optimal sensitivity of the initial test can be ‘weeded out’ by 12 lead ECG). The initial positives can be regarded as diagnostic in the presence of high specificity because high specificity means that almost all initial positives will have true AF. Because there are likely to be fewer initial positives than initial negatives, using a highly sensitive test is likely to lead to fewer people going on to the 12-lead test than use of a highly specific test. A highly sensitive initial test is therefore the preferable option for a two-stage process because the purpose of a two-stage process is to limit use of the resource-intensive 12 lead ECG.

Positive predictive value (PPV) and negative predictive value (NPV) are important for health economic considerations but are less important for evaluating clinical utility, and are often unreliable when calculated from study data as they are dependent on the prevalence which may not always be representative in studies. The aim had been to calculate PPV and NPV for any tools that had good evidence of adequate sensitivity and specificity in relation to an agreed prevalence rate of AF. However, this was not carried out because no tools were identified.

For the RCT review, outcomes were quality of life, mortality, stroke and thromboembolism, Major bleeding, all cause hospitalisation, confirmed diagnosis of AF and initiated anticoagulants for AF. All were regarded as critical by the committee, but quality of life, stroke/systemic embolism, mortality, and confirmed diagnosis of AF were deemed the most relevant for decision-making. These were prioritised over other critical outcomes because ‘quality of life’ was felt to provide the most comprehensive measure of benefit to the patient, ‘stroke and systemic thromboembolism’ was regarded as the major serious complication of AF, ‘mortality’ was felt to best characterise the harms of treatment, and ‘confirmed diagnosis of AF’ was thought to best characterise the benefits of treatment.

1.7.1.2. The quality of the evidence

For the diagnostic accuracy evidence, most data were rated as at serious or very serious risk of bias, because of a lack of simultaneity between index and reference tests, and because of a lack of blinding in some studies. Indirectness was also often rated as serious because the populations in studies differed from the protocol definition. Overall, most data were rated as low or very low. For the RCT evidence, a similar picture existed. Serious or very serious risk of bias was largely due to issues around selection and attrition bias, and again indirectness of populations was a major issue. Outcomes were therefore mostly rated as low or very low.

1.7.1.3. Benefits and harms

The diagnostic accuracy data for the different index test devices in relation to the gold standard of 12 lead ECG interpreted by a cardiologist/electrophysiologist were initially discussed. These devices included mobile ECG devices, HR monitors, blood pressure measurements, photoplethysmographic technique, pulse palpation, other ECG measures and 12 lead ECG not interpreted by an expert. The sensitivity and specificity of the majority of these devices were regarded by the committee as insufficiently high to permit their use as a single diagnostic test. Some devices, such as HR monitors, BP devices or plethysmographic devices, did approach 100% sensitivity and specificity, but these had often been tested in small samples leading to imprecise estimates. Alternatively, such estimates were from large but solitary studies. The committee noted that accuracy differed quite widely between different studies looking at the same test and they were therefore unable to make recommendations based on results from single studies.

Having decided that none of the tests could be used as an individual (definitive) diagnostic test, the committee discussed whether any of the tests could be used as a first-line test, prior to 12 lead ECG (please see ‘outcomes that matter the most’ above for an explanation of this process). The committee realised that such tests would need perfect or almost perfect sensitivity to avoid losing some people with AF from the diagnostic process (with enough specificity to allow a worthwhile reduction in the number going on to 12 lead testing compared to 12 lead testing used alone). The current recommendation is to use pulse palpation as the initial test, and thus an alternative test would need to have clear superiority in sensitivity over pulse palpation (with similar specificity) to justify replacement of pulse palpation. Some of the devices had sensitivity point estimates that exceeded those of pulse palpation, with upper 95% confidence intervals that extended closer to maximal sensitivity than those for pulse palpation. This provided weak evidence that some of the devices might be of greater use as a first line test than pulse palpation. However, the confidence intervals of the devices overlapped with those of pulse palpation, demonstrating a level of uncertainty about such superiority in the population. The committee were of the opinion that this level of uncertainty was insufficient to change the established practice of pulse palpation, which is a core clinical skill in widespread use, and which is extremely quick and low-cost to carry out. However, they felt that new devices had promise, which might be manifested in further high-quality research, and so a research recommendation was proposed, alongside a continuation of the current recommendation.

It is important to note a subtle change to the recommendations regarding the definitive test to be used if pulse irregularities are observed. In the previous guideline the recommendation had been to use ‘ECG’ as the definitive test, whereas in the present guideline we are specifying ‘12-lead ECG’ as the definitive test. This change was noted by the committee to be very important to prevent non-12 lead ECG such as lead I devices (which this review has shown to be lacking in adequate accuracy compared to 12 lead ECG) being used as the definitive test.

The diagnostic accuracy for the devices tested in relation to a longer-term gold standard (>24-hour ambulatory monitoring) were also considered by the committee. This evidence was regarded as particularly important as it was the only evidence able to inform the accuracy of detection of paroxysmal AF (12 lead ECG usually lasts only 10 seconds and so whilst it is perfectly good as a gold standard for detecting persistent AF it is often inadequate for detecting paroxysmal AF). The committee again noted that the evidence did not suggest that any specific test or device should be recommended but did note that the evidence clearly demonstrated that the accuracy of detection increased with the duration of testing. Therefore, the committee recommended that testing for suspected paroxysmal AF should be continued for as long as possible by any form of continuous or loop monitoring.

The committee agreed that the RCT review did not offer particularly useful evidence to inform recommendations, over and above the data provided by the diagnostic accuracy review. In particular, the committee highlighted that the follow up periods of the included studies were too short to allow a meaningful picture of downstream clinical outcomes. The RCT review was also noted to have serious gaps in terms of many of the available tests not having been studied.

1.7.2. Cost effectiveness and resource use

One cost-utility analysis was identified comparing single time point lead-I ECG devices with manual pulse palpation (MPP) followed by a 12-lead ECG in primary or secondary care for the detection of AF in people presenting to primary care with signs or symptoms of AF and who have an irregular pulse. This cost utility analysis was conducted as part of the NICE Diagnostic Guidance DG35 published in 2019 for lead-I devices. The study found that in all base case scenarios (these varied the time to and location of confirmatory 12 lead ECG) Kardia mobile, where treatment for AF is initiated following a positive result, ahead of confirmatory 12-lead ECG test, was the more cost-effective than the standard diagnostic pathway where no treatment is initiated until 12 lead ECG testing is complete. Furthermore, Kardia Mobile dominated (less costly and more effective) all other lead-I devices included in the analysis. This study was partially applicable as it did not include all comparators in the protocol for this question. There were potential serious limitations, primarily due to the fact the sensitivity and specificity data used in this analysis was from studies conducted in asymptomatic patients, and so this was indirect evidence. Furthermore, the economic evaluation is only relevant to primary care practices where patients have to wait at least 48 hours between an initial consultation with the GP and a 12-lead ECG.

In addition to this study, unit costs for different methods of detecting AF were presented, including current practice that is manual pulse palpitation followed by 12-lead ECG in those with an irregular pulse. The committee noted that although the lead-I devices do not appear particularly costly per use; they may add a significant resource burden in terms of the need for expert interpretation. This would either require training of GPs or would necessitate sending lead-I results to cardiologists for guidance and advice.

The committee considered the published health economic analysis alongside the clinical evidence and concluded that there was insufficient direct evidence to support replacing the current methods of detecting AF. In particular, the health economic evidence is based on indirect clinical evidence and there is uncertainty as to whether the sensitivity and specificity can be translated from an asymptomatic to a symptomatic AF population. This is in line with the guidance from DG35.

Overall, therefore the committee have kept the previous recommendations, only adjusting the wording to make these clearer. As they represent current practice, no resource impact is anticipated.

1.7.3. Other factors the committee took into account

The committee noted that the benefit of anticoagulation for asymptomatic AF that has not been documented on 12 lead ECG is uncertain and research is currently being conducted.

The committee noted that the use of hand-held devices could improve diagnosis in people who find it impossible or difficult to access EEG services, for example people in care homes.

The committee acknowledged the importance of primary care networks, including nurses and pharmacists, in the detection of AF in the community.

The committee highlighted that stroke prevention is one of the five programme work streams in the National Stroke Programme, which underpins the Long Term Plan with actions specifically around better diagnosis and management of AF. Integrated Stroke Delivery Networks have been set up across England to deliver on these commitments locally, and to implement improvements across the pathway at a regional level and should support efforts to improve AF detection and management.

The committee noted the challenges to delivery of healthcare in the context of COVID-19. Alternatives to face-face consultations should be explored with additional support to help people manage their condition. To mitigate against the current obstacles to in-person AF detection and management, NHS England and Improvement’s NHS at Home initiative, for example, aims to support people to remote monitor their health conditions and to use technology to allow clinicians to monitor their conditions remotely.

The committee noted that opportunistic screening was outside of the remit for this guideline.

References

1.: Acampa M, Lazzerini PE, Guideri F, Tassi R, Andreini I, Domenichelli C et al. Electrocardiographic predictors of silent atrial fibrillation in cryptogenic stroke. Heart, Lung & Circulation. 2019; 28(11):1664–1669 [PubMed: 30527848]

2.: Adami A, Gentile C, Hepp T, Molon G, Gigli GL, Valente M et al. Electrocardiographic RR interval dynamic analysis to identify acute stroke patients at high risk for atrial fibrillation episodes during stroke unit admission. Translational Stroke Research. 2019; 10(3):273–278 [PMC free article: PMC6526141] [PubMed: 29971705]

3.: Afzal MR, Gunda S, Waheed S, Sehar N, Maybrook RJ, Dawn B et al. Role of outpatient cardiac rhythm monitoring in cryptogenic stroke: a systematic review and meta-analysis. Pacing and Clinical Electrophysiology. 2015; 38(10):1236–1245 [PubMed: 26172621]

4.: Alshraideh H, Otoom M, Al-Araida A, Bawaneh H, Bravo J. A web based cardiovascular disease detection system. Journal of Medical Systems. 2015; 39(122) [PubMed: 26293754]

5.: Alves M, Narciso MR, Cruz J, Rocha M, Fonseca T. Paroxysmal atrial fibrillation detection in patients with acute ischemic stroke through prolonged Holter: prospective study. Aging Clinical and Experimental Research. 2019; 31(4):469–474 [PubMed: 30054893]

6.: Antonicelli R, Ripa C, Abbatecola AM, Capparuccia CA, Ferrara L, Spazzafumo L. Validation of the 3-lead tele-ECG versus the 12-lead tele-ECG and the conventional 12-lead ECG method in older people. Journal of Telemedicine and Telecare. 2012; 18(2):104–108 [PubMed: 22267307]

7.: Arevalo-Manso JJ, Martinez-Sanchez P, Fuentes B, Ruiz-Ares G, Sanz-Cuesta BE, Prefasi D et al. Can we improve the early detection of atrial fibrillation in a stroke unit? Detection rate of a monitor with integrated detection software. European Journal of Cardiovascular Nursing. 2016; 15(1):64–71 [PubMed: 25230856]

8.: Athif M, Yasawardene PC, Daluwatte C. Detecting atrial fibrillation from short single lead ECGs using statistical and morphological features. Physiological Measurement. 2018; 39(6):064002 [PubMed: 29767635]

9.: Attia ZI, Noseworthy PA, Lopez-Jimenez F, Asirvatham SJ, Deshmukh AJ, Gersh BJ et al. An artificial intelligence-enabled ECG algorithm for the identification of patients with atrial fibrillation during sinus rhythm: a retrospective analysis of outcome prediction. Lancet. 2019; 394(10201):861–867 [PubMed: 31378392]

10.: Baalman SWE, Mittal S, Boersma LVA, Perschbacher D, Brisben AJ, Mahajan D et al. Real-world performance of the atrial fibrillation monitor in patients with a subcutaneous ICD. Pacing and Clinical Electrophysiology. 2020; 10.1111/pace.14010 [PMC free article: PMC7754353] [PubMed: 32662101] [CrossRef]

11.: Barrett PM, Komatireddy R, Haaser S, Topol S, Sheard J, Encinas J et al. Comparison of 24-hour Holter monitoring with 14-day novel adhesive patch electrocardiographic monitoring. American Journal of Medicine. 2014; 127(1):95.e11–95.e17 [PMC free article: PMC3882198] [PubMed: 24384108]

12.: Barthelemy JC, Feasson-Gerard S, Garnier P, Gaspoz JM, Da Costa A, Michel D et al. Automatic cardiac event recorders reveal paroxysmal atrial fibrillation after unexplained strokes or transient ischemic attacks. Annals of Noninvasive Electrocardiology. 2003; 8(3):194–199 [PMC free article: PMC6932331] [PubMed: 14510653]

13.: Bell C, Kapral M. Use of ambulatory electrocardiography for the detection of paroxysmal atrial fibrillation in patients with stroke. Canadian Journal of Neurological Sciences. 2000; 27(1):25–31 [PubMed: 10676584]

14.: Berge T, Brynildsen J, Larssen HKN, Onarheim S, Jenssen GR, Ihle-Hansen H et al. Systematic screening for atrial fibrillation in a 65-year-old population with risk factors for stroke: data from the Akershus Cardiac Examination 1950 study. Europace: European Pacing, Arrhythmias, and Cardiac Electrophysiology. 2018; 20(FI3):f299–f305 [PubMed: 29095966]

15.: Berge T, Brynildsen J, Larssen HKN, Onarheim S, Jenssen GR, Ihle-Hansen H et al. Systematic screening for atrial fibrillation in 65-year-olds with risk factors for stroke. Data from the ACE 1950 Study. Europace: European Pacing, Arrhythmias, and Cardiac Electrophysiology. 2017; 19 (Suppl 3):iii355

16.: Bettin M, Dechering D, Kochhauser S, Bode N, Eckardt L, Frommeyer G et al. Extended ECG monitoring with an implantable loop recorder in patients with cryptogenic stroke: time schedule, reasons for explantation and incidental findings (results from the TRACK-AF trial). Clinical Research in Cardiology. 2019; 108(3):309–314 [PubMed: 30167809]

17.: Beukema R, Beukema WP, Sie HT, Misier AR, Delnoy PP, Elvan A. Monitoring of atrial fibrillation burden after surgical ablation: relevancy of end-point criteria after radiofrequency ablation treatment of patients with lone atrial fibrillation. Interactive Cardiovascular and Thoracic Surgery. 2009; 9(6):956–959 [PubMed: 19762419]

18.: Bonomi AG, Schipper F, Eerikainen LM, Margarito J, Van Dinther R, Muesch G et al. Atrial fibrillation detection using a novel cardiac ambulatory monitor based on photo-plethysmography at the wrist. Journal of the American Heart Association. 2018; 7(15):e009351 [PMC free article: PMC6201454] [PubMed: 30371247]

19.: Botto GL, Padeletti L, Santini M, Capucci A, Gulizia M, Zolezzi F et al. Presence and duration of atrial fibrillation detected by continuous monitoring: crucial implications for the risk of thromboembolic events. Journal of Cardiovascular Electrophysiology. 2009; 20(3):241–248 [PubMed: 19175849]

20.: Bourdillon PJ, Kilpatrick D. Clinicians, the Mount Sinai program and the Veterans’ Administration program evaluated against clinico-pathological data derived independently of the electrocardiogram. European Journal of Cardiology. 1978; 8(4–5):395–412 [PubMed: 152709]

21.: Brasier N, Raichle CJ, Dorr M, Becke A, Nohturfft V, Weber S et al. Detection of atrial fibrillation with a smartphone camera: first prospective, international, two-centre, clinical validation study (DETECT AF PRO). Europace: European Pacing, Arrhythmias, and Cardiac Electrophysiology. 2019; 21(1):41–47 [PMC free article: PMC6321964] [PubMed: 30085018]

22.: Brembilla-Perrot B, Luporsi JD, Louis S, Kaminsky P. Long-term follow-up of patients with myotonic dystrophy: an electrocardiogram every year is not necessary. Europace: European Pacing, Arrhythmias, and Cardiac Electrophysiology. 2011; 13(2):251–257 [PubMed: 21113052]

23.: Brito R, Mondouagne LP, Stettler C, Combescure C, Burri H. Automatic atrial fibrillation and flutter detection by a handheld ECG recorder, and utility of sequential finger and precordial recordings. Journal of Electrocardiology. 2018; 51(6):1135–1140 [PubMed: 30497745]

24.: Brown DL, Xu G, Belinky Krzyske AM, Buhay NC, Blaha M, Wang MM et al. Electrocardiomatrix Facilitates Accurate Detection of Atrial Fibrillation in Stroke Patients. Stroke. 2019; 50(7):1676–1681 [PubMed: 31177972]

25.: Buechi R, Faes L, Bachmann LM, Thiel MA, Bodmer NS, Schmid MK et al. Evidence assessing the diagnostic performance of medical smartphone apps: a systematic review and exploratory meta-analysis. BMJ Open. 2017; 7(12):e018280 [PMC free article: PMC5735404] [PubMed: 29247099]

26.: Bumgarner JM, Lambert CT, Hussein AA, Cantillon DJ, Baranowski B, Wolski K et al. Smartwatch algorithm for automated detection of atrial fibrillation. Journal of the American College of Cardiology. 2018; 71(21):2381–2388 [PubMed: 29535065]

27.: Burkowitz J, Merzenich C, Grassme K, Bruggenjurgen B. Insertable cardiac monitors in the diagnosis of syncope and the detection of atrial fibrillation: a systematic review and meta-analysis. European Journal of Preventive Cardiology. 2016; 23(12):1261–1272 [PubMed: 26864396]

28.: Busch MC, Gross S, Alte D, Kors JA, Volzke H, Ittermann T et al. Impact of atrial fibrillation detected by extended monitoring-a population-based cohort study. Annals of Noninvasive Electrocardiology. 2017; 22(6):e12453 [PMC free article: PMC6931841] [PubMed: 28440600]

29.: Caldwell JC, Borbas Z, Donald A, Clifford A, Bolger L, Black A et al. Simplified electrocardiogram sampling maintains high diagnostic capability for atrial fibrillation: implications for opportunistic atrial fibrillation screening in primary care. Europace: European Pacing, Arrhythmias, and Cardiac Electrophysiology. 2012; 14(2):191–196 [PubMed: 21993432]

30.: Callizo J, Feltgen N, Ammermann A, Ganser J, Bemme S, Bertelmann T et al. Atrial fibrillation in retinal vascular occlusion disease and non-arteritic anterior ischemic optic neuropathy. PloS One. 2017; 12(8):e0181766 [PMC free article: PMC5542383] [PubMed: 28771491]

31.: Censi F, Calcagnini G, Mattei E, Gargaro A, Biancalana G, Capucci A. Simulation of monitoring strategies for atrial arrhythmia detection. Annali dell’Istituto Superiore di Sanità. 2013; 49(2):176–182 [PubMed: 23771262]

32.: Chan PH, Wong CK, Poh YC, Pun L, Leung WW, Wong YF et al. Diagnostic performance of a smartphone-based photoplethysmographic application for atrial fibrillation screening in a primary care setting. Journal of the American Heart Association. 2016; 5(7):21 [PMC free article: PMC5015379] [PubMed: 27444506]

33.: Chan PH, Wong CK, Pun L, Wong YF, Wong MM, Chu DW et al. Diagnostic performance of an automatic blood pressure measurement device, Microlife WatchBP Home A, for atrial fibrillation screening in a real-world primary care setting. BMJ Open. 2017; 7(6):e013685 [PMC free article: PMC5577883] [PubMed: 28619766]

34.: Chan PH, Wong CK, Pun L, Wong YF, Wong MMY, Chu DWS et al. Head-to-head comparison of the AliveCor heart monitor and Microlife WatchBP Office AFIB for atrial fibrillation screening in a primary care setting. Circulation. 2017; 135(1):110–112 [PubMed: 28028066]

35.: Charitos EI, Stierle U, Ziegler PD, Baldewig M, Robinson DR, Sievers HH et al. A comprehensive evaluation of rhythm monitoring strategies for the detection of atrial fibrillation recurrence: insights from 647 continuously monitored patients and implications for monitoring after therapeutic interventions. Circulation. 2012; 126(7):806–814 [PubMed: 22824434]

36.: Chen E, Jiang J, Su R, Gao M, Zhu S, Zhou J et al. A new smart wristband equipped with an artificial intelligence algorithm to detect atrial fibrillation. Heart Rhythm. 2020; 17(5 Pt B):847–853 [PubMed: 32354449]

37.: Chen YH, Hung CS, Huang CC, Hung YC, Hwang JJ, Ho YL. Atrial Fibrillation Screening in Nonmetropolitan Areas Using a Telehealth Surveillance System With an Embedded Cloud-Computing Algorithm: Prospective Pilot Study. JMIR MHealth and UHealth. 2017; 5(9):e135 [PMC free article: PMC5635230] [PubMed: 28951384]

38.: Choe WC, Passman RS, Brachmann J, Morillo CA, Sanna T, Bernstein RA et al. A Comparison of Atrial Fibrillation Monitoring Strategies After Cryptogenic Stroke (from the Cryptogenic Stroke and Underlying AF Trial). American Journal of Cardiology. 2015; 116(6):890–893 [PubMed: 26183793]

39.: Chong JW, Cho CH, Tabei F, Le-Anh D, Esa N, McManus DD et al. Motion and noise artifact-resilient atrial fibrillation detection using a smartphone. IEEE Journal on Emerging and Selected Topics in Circuits & Systems. 2018; 8(2):230–239 [PMC free article: PMC6345530] [PubMed: 30687580]

40.: Chong JW, Esa N, McManus DD, Chon KH. Arrhythmia discrimination using a smart phone. IEEE journal of biomedical and health informatics. 2015; 19(3):815–824 [PMC free article: PMC6599713] [PubMed: 25838530]

41.: Chovancik J, Bulkova V, Wichterle D, Toman O, Rybka L, Januska J et al. Comparison of two modes of long-term ECG monitoring to assess the efficacy of catheter ablation for paroxysmal atrial fibrillation. Biomedical Papers of the Medical Faculty of Palacky University in Olomouc, Czech Republic. 2019; 163(1):54–60 [PubMed: 29955186]

42.: Christensen LM, Krieger DW, Hojberg S, Pedersen OD, Karlsen FM, Jacobsen MD et al. Paroxysmal atrial fibrillation occurs often in cryptogenic ischaemic stroke. Final results from the SURPRISE study. European Journal of Neurology. 2014; 21(6):884–889 [PubMed: 24628954]

43.: Ciconte G, Saviano M, Giannelli L, Calovic Z, Baldi M, Ciaccio C et al. Atrial fibrillation detection using a novel three-vector cardiac implantable monitor: the atrial fibrillation detect study. Europace: European Pacing, Arrhythmias, and Cardiac Electrophysiology. 2017; 19(7):1101–1108 [PubMed: 27702865]

44.: Conroy T, Guzman JH, Hall B, Tsouri G, Couderc JP. Detection of atrial fibrillation using an earlobe photoplethysmographic sensor. Physiological Measurement. 2017; 38(10):1906–1918 [PubMed: 28836507]

45.: Cooke G, Doust J, Sanders S. Is pulse palpation helpful in detecting atrial fibrillation? A systematic review. Journal of Family Practice. 2006; 55(2):130–134 [PubMed: 16451780]

46.: Couderc JP, Kyal S, Mestha LK, Xu B, Peterson DR, Xia X et al. Detection of atrial fibrillation using contactless facial video monitoring. Heart Rhythm. 2015; 12(1):195–201 [PubMed: 25179488]

47.: Coutts SB, Choi PMC. Seven days of non-invasive cardiac monitoring early postischaemic stroke or TIA increases atrial fibrillation detection rate compared with current guideline-based practice. Evidence-Based Medicine. 2014; 19(4):152 [PubMed: 24284313]

48.: Cuadrado-Godia E, Benito B, Ois A, Valles E, Rodriguez-Campello A, Giralt-Steinhauer E et al. Ultra-early continuous cardiac monitoring improves atrial fibrillation detection and prognosis of patients with cryptogenic stroke. European Journal of Neurology. 2019; 27(2):244–250 [PubMed: 31424609]

49.: Cunha S, Antunes E, Antoniou S, Tiago S, Relvas R, Fernandez-Llimos F et al. Raising awareness and early detection of atrial fibrillation, an experience resorting to mobile technology centred on informed individuals. Research In Social and Administrative Pharmacy. 2019; 16(6):787–792 [PubMed: 31473110]

50.: Czabanski R, Horoba K, Wrobel J, Matonia A, Martinek R, Kupka T et al. Detection of atrial fibrillation episodes in long-term heart rhythm signals using a support vector machine. Sensors. 2020; 20(3):30 [PMC free article: PMC7038413] [PubMed: 32019220]

51.: Dagres N, Kottkamp H, Piorkowski C, Weis S, Arya A, Sommer P et al. Influence of the duration of Holter monitoring on the detection of arrhythmia recurrences after catheter ablation of atrial fibrillation implications for patient follow-up. International Journal of Cardiology. 2010; 139(3):305–306 [PubMed: 18990460]

52.: Damiano RJ, Jr., Lawrance CP, Saint LL, Henn MC, Sinn LA, Kruse J et al. Detection of atrial fibrillation after surgical ablation: conventional versus continuous monitoring. Annals of Thoracic Surgery. 2016; 101(1):42–47 [PMC free article: PMC5519145] [PubMed: 26507426]

53.: De Lucia R, Zucchelli G, Barletta V, Di Cori A, Giannotti Santoro M, Parollo M et al. The in-ear region as a novel anatomical site for ECG signal detection: validation study on healthy volunteers. Journal of Interventional Cardiac Electrophysiology. 2020; 10.1007/s10840-020-00709-x [PubMed: 32064554] [CrossRef]

54.: de Voogt WG, van Hemel NM, van de Bos AA, Koistinen J, Fast JH. Verification of pacemaker automatic mode switching for the detection of atrial fibrillation and atrial tachycardia with Holter recording. Europace: European Pacing, Arrhythmias, and Cardiac Electrophysiology. 2006; 8(11):950–961 [PubMed: 17043069]

55.: DeBoard Z, Doty JR. Evaluation of new generation loop recorders placed during surgical ablation for atrial fibrillation. Journal of Cardiac Surgery. 2018; 33(7):416–419 [PubMed: 29911345]

56.: Defaye P, Dournaux F, Mouton E. Prevalence of supraventricular arrhythmias from the automated analysis of data stored in the DDD pacemakers of 617 patients: the AIDA study. The AIDA Multicenter Study Group. Automatic Interpretation for Diagnosis Assistance. Pacing and Clinical Electrophysiology. 1998; 21(1 Pt 2):250–255 [PubMed: 9474682]

57.: Derkac WM, Finkelmeier JR, Horgan DJ, Hutchinson MD. Diagnostic yield of asymptomatic arrhythmias detected by mobile cardiac outpatient telemetry and autotrigger looping event cardiac monitors. Journal of Cardiovascular Electrophysiology. 2017; 28(12):1475–1478 [PubMed: 28940881]

58.: Desteghe L, Raymaekers Z, Lutin M, Vijgen J, Dilling-Boer D, Koopman P et al. Performance of handheld electrocardiogram devices to detect atrial fibrillation in a cardiology and geriatric ward setting. Europace: European Pacing, Arrhythmias, and Cardiac Electrophysiology. 2017; 19(1):29–39 [PubMed: 26893496]

59.: Diamantino AC, Nascimento BR, Beaton AZ, Nunes MCP, Oliveira KKB, Rabelo LC et al. Atrial fibrillation detection with a portable device during cardiovascular screening in primary care. Heart. 2020; 106(16):1261–1266 [PubMed: 32019822]

60.: Diamantopoulos A, Sawyer LM, Lip GY, Witte KK, Reynolds MR, Fauchier L et al. Cost-effectiveness of an insertable cardiac monitor to detect atrial fibrillation in patients with cryptogenic stroke. International Journal of Stroke. 2016; 11(3):302–312 [PubMed: 26763916]

61.: Dimarco AD, Onwordi EN, Murphy CF, Walters EJ, Willis L, Mullan NJ et al. Diagnostic utility of real-time smartphone ECG in the initial investigation of palpitations. British Journal of Cardiology. 2018; 10.5837/bjc.2018.006 [CrossRef]

62.: Ding EY, Albuquerque D, Winter M, Binici S, Piche J, Bashar SK et al. Novel method of atrial fibrillation case identification and burden estimation using the MIMIC-III electronic health data set. Journal of Intensive Care Medicine. 2019; 34(10):851–857 [PMC free article: PMC7050656] [PubMed: 31354020]

63.: Doliwa PS, Frykman V, Rosenqvist M. Short-term ECG for out of hospital detection of silent atrial fibrillation episodes. Scandinavian Cardiovascular Journal. 2009; 43(3):163–168 [PubMed: 19096977]

64.: Dorr M, Nohturfft V, Brasier N, Bosshard E, Djurdjevic A, Gross S et al. The WATCH AF Trial: SmartWATCHes for detection of atrial fibrillation. JACC: Clinical Electrophysiology. 2019; 5(2):199–208 [PubMed: 30784691]

65.: Duarte R, Stainthorpe A, Greenhalgh J, Richardson M, Nevitt S, Mahon J et al. Lead-I ECG for detecting atrial fibrillation in patients with an irregular pulse using single time point testing: a systematic review and economic evaluation. Health Technology Assessment (Winchester, England). 2020; 24(3):1–164 [PMC free article: PMC6983912] [PubMed: 31933471]

66.: Dussault C, Toeg H, Nathan M, Wang ZJ, Roux JF, Secemsky E. Electrocardiographic monitoring for detecting atrial fibrillation after ischemic stroke or transient ischemic attack. Circulation: Arrhythmia and Electrophysiology. 2015; 8(2):263–269 [PubMed: 25639643]

67.: Edgerton JR, Mahoney C, Mack MJ, Roper K, Herbert MA. Long-term monitoring after surgical ablation for atrial fibrillation: How much is enough? Journal of Thoracic and Cardiovascular Surgery. 2011; 142(1):162–165 [PubMed: 21377697]

68.: Eitel C, Husser D, Hindricks G, Fruhauf M, Hilbert S, Arya A et al. Performance of an implantable automatic atrial fibrillation detection device: impact of software adjustments and relevance of manual episode analysis. Europace: European Pacing, Arrhythmias, and Cardiac Electrophysiology. 2011; 13(4):480–485 [PMC free article: PMC3065917] [PubMed: 21325346]

69.: Elijovich L, Josephson SA, Fung GL, Smith WS. Intermittent atrial fibrillation may account for a large proportion of otherwise cryptogenic stroke: a study of 30-day cardiac event monitors. Journal of Stroke and Cerebrovascular Diseases. 2009; 18(3):185–189 [PubMed: 19426887]

70.: Engdahl J, Andersson L, Mirskaya M, Rosenqvist M. Stepwise screening of atrial fibrillation in a 75-year-old population: implications for stroke prevention. Circulation. 2013; 127(8):930–937 [PubMed: 23343564]

71.: Engdahl J, Holmen A, Rosenqvist M, Stromberg U. A prospective 5-year follow-up after population-based systematic screening for atrial fibrillation. Europace: European Pacing, Arrhythmias, and Cardiac Electrophysiology. 2018; 20(FI_3):f306–f311 [PubMed: 29688312]

72.: Ermini G, Filippi A, Salera M. Switching from traditional to automatic sphygmomanometer increases opportunistic detection of atrial fibrillation in hypertensive patients. British Journal of Medical Practitioners. 2013; 6(1):A616

73.: Etiwy M, Akhrass Z, Gillinov L, Alashi A, Wang R, Blackburn G et al. Accuracy of wearable heart rate monitors in cardiac rehabilitation. Cardiovascular Diagnosis and Therapy. 2019; 9(3):262–271 [PMC free article: PMC6603497] [PubMed: 31275816]

74.: Evans GF, Shirk A, Muturi P, Soliman EZ. Feasibility of using mobile ECG recording technology to detect atrial fibrillation in low-resource settings. Global heart. 2017; 12(4):285–289 [PubMed: 28302547]

75.: Eysenck W, Ammar A, Kanthasamy V, Freemantle N, Veasey RA, Patel NR et al. A trial of three non-invasive blood pressure monitors compared with invasive blood pressure assessment in atrial fibrillation and sinus rhythm. International Journal of Clinical Practice. 2019:e13410 [PubMed: 31464020]

76.: Fallet S, Lemay M, Renevey P, Leupi C, Pruvot E, Vesin JM. Can one detect atrial fibrillation using a wrist-type photoplethysmographic device? Medical and Biological Engineering and Computing. 2019; 57(2):477–487 [PubMed: 30218408]

77.: Fan YY, Li YG, Li J, Cheng WK, Shan ZL, Wang YT et al. Diagnostic performance of a smart device with photoplethysmography technology for atrial fibrillation detection: pilot study (Pre-mAFA II Registry). JMIR MHealth and UHealth. 2019; 7(3):e11437 [PMC free article: PMC6423467] [PubMed: 30835243]

78.: Gaillard N, Deltour S, Vilotijevic B, Hornych A, Crozier S, Leger A et al. Detection of paroxysmal atrial fibrillation with transtelephonic EKG in TIA or stroke patients. Neurology. 2010; 74(21):1666–1670 [PubMed: 20498434]

79.: Gandolfo C, Balestrino M, Bruno C, Finocchi C, Reale N. Validation of a simple method for atrial fibrillation screening in patients with stroke. Neurological Sciences. 2015; 36(9):1675–1678 [PubMed: 25926072]

80.: Ghazal F, Theobald H, Rosenqvist M, Al-Khalili F. Validity of daily self-pulse palpation for atrial fibrillation screening in patients 65 years and older: a cross-sectional study. PLoS Medicine. 2020; 17(3):e1003063 [PMC free article: PMC7108684] [PubMed: 32231369]

81.: Godin R, Yeung C, Baranchuk A, Guerra P, Healey JS. Screening for atrial fibrillation using a mobile, single-lead electrocardiogram in canadian primary care clinics. Canadian Journal of Cardiology. 2019; 35(7):840–845 [PubMed: 31292082]

82.: Gregg RE, Zhou SH, Lindauer JM, Feild DQ, Helfenbein ED. Where do derived precordial leads fail? Journal of Electrocardiology. 2008; 41(6):546–552 [PubMed: 18817921]

83.: Grond M, Jauss M, Hamann G, Stark E, Veltkamp R, Nabavi D et al. Improved detection of silent atrial fibrillation using 72-hour holter ecg in patients with ischemic stroke: A prospective multicenter cohort study. Stroke. 2013; 44(12):3357–3364 [PubMed: 24130137]

84.: Groschel S, Lange B, Grond M, Jauss M, Kirchhof P, Rostock T et al. Automatic Holter electrocardiogram analysis in ischaemic stroke patients to detect paroxysmal atrial fibrillation: ready to replace physicians? European Journal of Neurology. 2020; 27(7):1272–1278 [PubMed: 32279383]

85.: Groschel S, Lange B, Wasser K, Hahn M, Wachter R, Groschel K et al. Software-based analysis of 1-hour Holter ECG to select for prolonged ECG monitoring after stroke. Annals of Clinical & Translational Neurology. 2020; 7:1779–1787 [PMC free article: PMC7545589] [PubMed: 32862499]

86.: Guan J, Wang A, Song W, Obore N, He P, Fan S et al. Screening for arrhythmia with the new portable single-lead electrocardiographic device (SnapECG): an application study in community-based elderly population in Nanjing, China. Aging Clinical and Experimental Research. 2020; 10.1007/s40520-020-01512-4 [PubMed: 32144732] [CrossRef]

87.: Gunalp M, Atalar E, Coskun F, Yilmaz A, Aksoyek S, Aksu NM et al. Holter monitoring for 24 hours in patients with thromboembolic stroke and sinus rhythm diagnosed in the emergency department. Advances in Therapy. 2006; 23(6):854–860 [PubMed: 17276953]

88.: Guo Y, Wang H, Zhang H, Liu T, Liang Z, Xia Y et al. Mobile Photoplethysmographic Technology to Detect Atrial Fibrillation. Journal of the American College of Cardiology. 2019; 74(29):2365–2375 [PubMed: 31487545]

89.: Guo Y, Wang H, Zhang H, Liu T, Liang Z, Xia Y et al. Mobile photoplethysmographic technology to detect atrial fibrillation. Journal of the American College of Cardiology. 2019; 74(19):2365–2375 [PubMed: 31487545]

90.: Haberman ZC, Jahn RT, Bose R, Tun H, Shinbane JS, Doshi RN et al. Wireless smartphone ECG enables large-scale screening in diverse populations. Journal of Cardiovascular Electrophysiology. 2015; 26(5):520–526 [PubMed: 25651872]

91.: Hald J, Poulsen PB, Qvist I, Holm L, Wedell-Wedellsborg D, Dybro L et al. Opportunistic screening for atrial fibrillation in a real-life setting in general practice in Denmark-The Atrial Fibrillation Found On Routine Detection (AFFORD) non-interventional study. PloS One. 2017; 12(11):e0188086 [PMC free article: PMC5683635] [PubMed: 29131836]

92.: Hanke T, Charitos EI, Stierle U, Karluss A, Kraatz E, Graf B et al. Twenty-four-hour holter monitor follow-up does not provide accurate heart rhythm status after surgical atrial fibrillation ablation therapy: up to 12 months experience with a novel permanently implantable heart rhythm monitor device. Circulation. 2009; 120:(11 Suppl):S177–184 [PubMed: 19752365]

93.: Harju J, Tarniceriu A, Parak J, Vehkaoja A, Yli-Hankala A, Korhonen I. Monitoring of heart rate and inter-beat intervals with wrist plethysmography in patients with atrial fibrillation. Physiological Measurement. 2018; 39(6):065007 [PubMed: 29856730]

94.: Harris K, Edwards D, Mant J. How can we best detect atrial fibrillation? The journal of the Royal College of Physicians of Edinburgh. 2012; 42:(Suppl 18):5–22 [PubMed: 22518390]

95.: Hartikainen S, Lipponen JA, Hiltunen P, Rissanen TT, Kolk I, Tarvainen MP et al. Effectiveness of the chest strap electrocardiogram to detect atrial fibrillation. American Journal of Cardiology. 2019; 123(10):1643–1648 [PubMed: 30878151]

96.: Haverkamp HT, Fosse SO, Schuster P. Accuracy and usability of single-lead ECG from smartphones - a clinical study. Indian Pacing and Electrophysiology Journal. 2019; 19(4):145–149 [PMC free article: PMC6697525] [PubMed: 30794928]

97.: Hendrikx T, Rosenqvist M, Wester P, Sandstrom H, Hornsten R. Intermittent short ECG recording is more effective than 24-hour Holter ECG in detection of arrhythmias. BMC Cardiovascular Disorders. 2014; 14:41 [PMC free article: PMC4234325] [PubMed: 24690488]

98.: Hickey KT, Biviano AB, Garan H, Sciacca RR, Riga T, Warren K et al. Evaluating the utility of mHealth ECG heart monitoring for the detection and management of atrial fibrillation in clinical practice. Journal of Atrial Fibrillation. 2017; 9(5):1546 [PMC free article: PMC5673393] [PubMed: 29250277]

99.: Higgins P, Dawson J, Macfarlane PW, McArthur K, Langhorne P, Lees KR. Predictive value of newly detected atrial fibrillation paroxysms in patients with acute ischemic stroke, for atrial fibrillation after 90 days. Stroke. 2014; 45(7):2134–2136 [PubMed: 24938848]

100.: Higgins P, MacFarlane PW, Dawson J, McInnes GT, Langhorne P, Lees K. ECG monitoring strategy to identify AF after stroke. ISRCTN Registry. Springer Nature, 2010. Available from: 10.1002/central/CN-00983010/full [CrossRef]

101.: Himmelreich JCL, Karregat EPM, Lucassen WAM, van Weert HCPM, de Groot JR, Handoko ML et al. Diagnostic accuracy of a smartphone-operated, single-lead electrocardiography device for detection of rhythm and conduction abnormalities in primary care. Annals of Family Medicine. 2019; 17(5):403–411 [PMC free article: PMC7032908] [PubMed: 31501201]

102.: Hindricks G, Pokushalov E, Urban L, Taborsky M, Kuck KH, Lebedev D et al. Performance of a new leadless implantable cardiac monitor in detecting and quantifying atrial fibrillation: Results of the XPECT trial. Circulation: Arrhythmia and Electrophysiology. 2010; 3(2):141–147 [PubMed: 20160169]

103.: Hisazaki K, Miyazaki S, Hasegawa K, Kaseno K, Amaya N, Shiomi Y et al. The P wave morphology in lead V7 on the synthesized 18-lead ECG is a useful parameter for identifying arrhythmias originating from the right inferior pulmonary vein. Heart and Vessels. 2019; 35(2):246–251 [PubMed: 31440830]

104.: Hobbs FD, Fitzmaurice DA, Mant J, Murray E, Jowett S, Bryan S et al. A randomised controlled trial and cost-effectiveness study of systematic screening (targeted and total population screening) versus routine practice for the detection of atrial fibrillation in people aged 65 and over. The SAFE study. Health Technology Assessment. 2005; 9(40) [PubMed: 16202350]

105.: Hochstadt A, Chorin E, Viskin S, Schwartz AL, Lubman N, Rosso R. Continuous heart rate monitoring for automatic detection of atrial fibrillation with novel bio-sensing technology. Journal of Electrocardiology. 2019; 52:23–27 [PubMed: 30476634]

106.: Inui T, Kohno H, Kawasaki Y, Matsuura K, Ueda H, Tamura Y et al. Use of a Smart Watch for Early Detection of Paroxysmal Atrial Fibrillation: Validation Study. JMIR Cardio. 2020; 4(1):e14857 [PMC free article: PMC7003123] [PubMed: 32012044]

107.: Ip JE. Wearable devices for cardiac rhythm diagnosis and management. JAMA. 2019; 321(4):337–338 [PubMed: 30633301]

108.: Ip JH, Viqar-Syed M, Grimes D, Xie Y, Jager K, Boak J et al. Surveillance of AF recurrence post-surgical AF ablation using implantable cardiac monitor. Journal of Interventional Cardiac Electrophysiology. 2012; 33(1):77–83 [PubMed: 21814825]

109.: Israel C, Kitsiou A, Kalyani M, Deelawar S, Ejangue LE, Rogalewski A et al. Detection of atrial fibrillation in patients with embolic stroke of undetermined source by prolonged monitoring with implantable loop recorders. Thrombosis and Haemostasis. 2017; 117(10):1962–1969 [PubMed: 28862284]

110.: Israel CW, Hugl B, Unterberg C, Lawo T, Kennis I, Hettrick D et al. Pace-termination and pacing for prevention of atrial tachyarrhythmias: results from a multicenter study with an implantable device for atrial therapy. Journal of Cardiovascular Electrophysiology. 2001; 12(10):1121–1128 [PubMed: 11699520]

111.: Jabaudon D, Sztajzel J, Sievert K, Landis T, Sztajzel R. Usefulness of ambulatory 7-day ECG monitoring for the detection of atrial fibrillation and flutter after acute stroke and transient ischemic attack. Stroke. 2004; 35(7):1647–1651 [PubMed: 15155965]

112.: Jacobs MS, Kaasenbrood F, Postma MJ, van Hulst M, Tieleman RG. Cost-effectiveness of screening for atrial fibrillation in primary care with a handheld, single-lead electrocardiogram device in the Netherlands. Europace: European Pacing, Arrhythmias, and Cardiac Electrophysiology. 2018; 20(1):12–18 [PubMed: 27733465]

113.: Jiang K, Huang C, Ye SM, Chen H. High accuracy in automatic detection of atrial fibrillation for Holter monitoring. Journal of Zhejiang University: Science B. 2012; 13(9):751–756 [PMC free article: PMC3437373] [PubMed: 22949366]

114.: Kaasenbrood F, Hollander M, Rutten FH, Gerhards LJ, Hoes AW, Tieleman RG. Yield of screening for atrial fibrillation in primary care with a hand-held, single-lead electrocardiogram device during influenza vaccination. Europace: European Pacing, Arrhythmias, and Cardiac Electrophysiology. 2016; 18(10):1514–1520 [PMC free article: PMC5072135] [PubMed: 26851813]

115.: Kabutoya T, Imai Y, Hoshide S, Kario K. Diagnostic accuracy of a new algorithm to detect atrial fibrillation in a home blood pressure monitor. Journal of Clinical Hypertension. 2017; 19(11):1143–1147 [PMC free article: PMC8030935] [PubMed: 28861938]

116.: Kabutoya T, Takahashi S, Watanabe T, Imai Y, Uemoto K, Yasui N et al. Diagnostic accuracy of an algorithm for detecting atrial fibrillation in a wrist-type pulse wave monitor. Journal of Clinical Hypertension. 2019; 21(9):1393–1398 [PMC free article: PMC8030341] [PubMed: 31420946]

117.: Kaleschke G, Hoffmann B, Drewitz I, Steinbeck G, Naebauer M, Goette A et al. Prospective, multicentre validation of a simple, patient-operated electrocardiographic system for the detection of arrhythmias and electrocardiographic changes. Europace: European Pacing, Arrhythmias, and Cardiac Electrophysiology. 2009; 11(10):1362–1368 [PubMed: 19797150]

118.: Kalidas V, Tamil LS. Detection of atrial fibrillation using discrete-state Markov models and Random Forests. Computers in Biology and Medicine. 2019; 113:103386 [PubMed: 31446318]

119.: Kallmunzer B, Bobinger T, Kahl N, Kopp M, Kurka N, Hilz MJ et al. Peripheral pulse measurement after ischemic stroke: a feasibility study. Neurology. 2014; 83(7):598–603 [PubMed: 25056581]

120.: Kallmunzer B, Breuer L, Hering C, Raaz-Schrauder D, Kollmar R, Huttner HB et al. A structured reading algorithm improves telemetric detection of atrial fibrillation after acute ischemic stroke. Stroke. 2012; 43(4):994–999 [PubMed: 22308240]

121.: Kane SA, Blake JR, McArdle FJ, Langley P, Sims AJ. Opportunistic detection of atrial fibrillation using blood pressure monitors: a systematic review. Open Heart. 2016; 3(1):e000362 [PMC free article: PMC4836305] [PubMed: 27099760]

122.: Kang SH, Joe B, Yoon Y, Cho GY, Shin I, Suh JW. Cardiac auscultation using smartphones: pilot study. JMIR MHealth and UHealth. 2018; 6(2):e49 [PMC free article: PMC5853766] [PubMed: 29490899]

123.: Kao WF, Hou SK, Huang CY, Chao CC, Cheng CC, Chen YJ. Assessment of the clinical efficacy of the heart spectrum blood pressure monitor for diagnosis of atrial fibrillation: An unblinded clinical trial. PloS One. 2018; 13(6):e0198852 [PMC free article: PMC6001975] [PubMed: 29902218]

124.: Karaoguz MR, Yurtseven E, Aslan G, Deliormanli BG, Adiguzel O, Gonen M et al. The quality of ECG data acquisition, and diagnostic performance of a novel adhesive patch for ambulatory cardiac rhythm monitoring in arrhythmia detection. Journal of Electrocardiology. 2019; 54:28–35 [PubMed: 30851474]

125.: Karregat EPM, Himmelreich JCL, Lucassen WAM, Busschers WB, van Weert H, Harskamp RE. Evaluation of general practitioners’ single-lead electrocardiogram interpretation skills: a case-vignette study. Family Practice. 2020; 10.1093/fampra/cmaa076 [PMC free article: PMC8006764] [PubMed: 32766703] [CrossRef]

126.: Karunadas CP, Mathew C. Comparison of arrhythmia detection by conventional Holter and a novel ambulatory ECG system using patch and Android App, over 24 h period. Indian Pacing and Electrophysiology Journal. 2020; 20(2):49–53 [PMC free article: PMC7082686] [PubMed: 31866554]

127.: Kashiwa A, Koyama F, Miyamoto K, Kamakura T, Wada M, Yamagata K et al. Performance of an atrial fibrillation detection algorithm using continuous pulse wave monitoring. Annals of Noninvasive Electrocardiology. 2019; 24(2):e12615 [PMC free article: PMC6931792] [PubMed: 30387545]

128.: Kearley K, Selwood M, Van Den Bruel A, Thompson M, Mant D, Hobbs FDR et al. Triage tests for identifying atrial fibrillation in primary care: a diagnostic accuracy study comparing single-lead ECG and modified BP monitors. BMJ Open. 2014; 4(5):e004565 [PMC free article: PMC4025411] [PubMed: 24793250]

129.: Kim NR, Choi CK, Kim HS, Oh SH, Yang JH, Lee KH et al. Screening for atrial fibrillation using a smartphone-based electrocardiogram in Korean elderly. Chonnam Medical Journal. 2020; 56(1):50–54 [PMC free article: PMC6976762] [PubMed: 32021842]

130.: Kircher S, Hindricks G, Sommer P. Long-term success and follow-up after atrial fibrillation ablation. Current Cardiology Reviews. 2012; 8(4):354–361 [PMC free article: PMC3492818] [PubMed: 22920479]

131.: Kishore A, Vail A, Majid A, Dawson J, Lees KR, Tyrrell PJ et al. Detection of atrial fibrillation after ischemic stroke or transient ischemic attack: a systematic review and meta-analysis. Stroke. 2014; 45(2):520–526 [PubMed: 24385275]

132.: Kollias A, Destounis A, Kalogeropoulos P, Kyriakoulis KG, Ntineri A, Stergiou GS. Atrial fibrillation detection during 24-hour ambulatory blood pressure monitoring: comparison with 24-hour electrocardiography. Hypertension. 2018; 72(1):110–115 [PubMed: 29735633]

133.: Koltowski L, Balsam P, Gllowczynska R, Rokicki JK, Peller M, Maksym J et al. Kardia Mobile applicability in clinical practice: a comparison of Kardia Mobile and standard 12-lead electrocardiogram records in 100 consecutive patients of a tertiary cardiovascular care center. Cardiology Journal. 2019; 10.5603/CJ.a2019.0001 [PMC free article: PMC8276994] [PubMed: 30644079] [CrossRef]

134.: Kong D, Zhu J, Wu S, Duan C, Lu L, Chen D. A novel IRBF-RVM model for diagnosis of atrial fibrillation. Computer Methods and Programs in Biomedicine. 2019; 177:183–192 [PubMed: 31319947]

135.: Korompoki E, Del Giudice A, Hillmann S, Malzahn U, Gladstone DJ, Heuschmann P et al. Cardiac monitoring for detection of atrial fibrillation after TIA: A systematic review and meta-analysis. International Journal of Stroke. 2017; 12(1):33–45 [PubMed: 27681890]

136.: Koshy AN, Sajeev JK, Nerlekar N, Brown AJ, Rajakariar K, Zureik M et al. Smart watches for heart rate assessment in atrial arrhythmias. International Journal of Cardiology. 2018; 266:124–127 [PubMed: 29887428]

137.: Koshy AN, Sajeev JK, Nerlekar N, Brown AJ, Rajakariar K, Zureik M et al. Utility of photoplethysmography for heart rate estimation among inpatients. Internal Medicine Journal. 2018; 48(5):587–591 [PubMed: 29722189]

138.: Kristensen AN, Jeyam B, Riahi S, Jensen MB. The use of a portable three-lead ECG monitor to detect atrial fibrillation in general practice. Scandinavian Journal of Primary Health Care. 2016; 34(3):304–308 [PMC free article: PMC5036021] [PubMed: 27409151]

139.: Krivoshei L, Weber S, Burkard T, Maseli A, Brasier N, Kuhne M et al. Smart detection of atrial fibrillation. Europace: European Pacing, Arrhythmias, and Cardiac Electrophysiology. 2017; 19(5):753–757 [PMC free article: PMC5437701] [PubMed: 27371660]

140.: Kvist LM, Vinter N, Urbonaviciene G, Lindholt JS, Diederichsen ACP, Frost L. Diagnostic accuracies of screening for atrial fibrillation by cardiac nurses versus radiographers. Open Heart. 2019; 6(1):e000942 [PMC free article: PMC6443120] [PubMed: 30997131]

141.: Kwon S, Hong J, Choi EK, Lee B, Baik C, Lee E et al. Detection of Atrial Fibrillation Using a Ring-Type Wearable Device (CardioTracker) and Deep Learning Analysis of Photoplethysmography Signals: Prospective Observational Proof-of-Concept Study. Journal of Medical Internet Research. 2020; 22(5):e16443 [PMC free article: PMC7273241] [PubMed: 32348254]

142.: Kwon S, Hong J, Choi EK, Lee E, Hostallero DE, Kang WJ et al. Deep learning approaches to detect atrial fibrillation using photoplethysmographic signals: Algorithms development study. JMIR MHealth and UHealth. 2019; 7(6):e12770 [PMC free article: PMC6592499] [PubMed: 31199302]

143.: Lahdenoja O, Hurnanen T, Iftikhar Z, Nieminen S, Knuutila T, Saraste A et al. Atrial fibrillation detection via accelerometer and gyroscope of a smartphone. IEEE journal of biomedical and health informatics. 2018; 22(1):108–118 [PubMed: 28391210]

144.: Lai D, Bu Y, Su Y, Zhang X, Ma CS. Non-Standardized Patch-Based ECG Lead Together With Deep Learning Based Algorithm for Automatic Screening of Atrial Fibrillation. IEEE Journal of Biomedical & Health Informatics. 2020; 24(6):1569–1578 [PubMed: 32175879]

145.: Langley P, Dewhurst M, Di Marco LY, Adams P, Dewhurst F, Mwita JC et al. Accuracy of algorithms for detection of atrial fibrillation from short duration beat interval recordings. Medical Engineering and Physics. 2012; 34(10):1441–1447 [PubMed: 22398415]

146.: Lau J, Lowres N, Neubeck L, Brieger D, Sy R, Galloway C et al. Performance of an automated iPhone ECG algorithm to diagnose atrial fibrillation in a community AF screening program (search-AF). Heart Lung and Circulation. 2013; 22:(Suppl 1):S205

147.: Lauschke J, Busch M, Haverkamp W, Bulava A, Schneider R, Andresen D et al. New implantable cardiac monitor with three-lead ECG and active noise detection. Herz. 2017; 42(6):585–592 [PubMed: 27796409]

148.: Lee J, Nam Y, McManus DD, Chon KH. Time-varying coherence function for atrial fibrillation detection. IEEE Transactions on Biomedical Engineering. 2013; 60(10):2783–2793 [PubMed: 23708769]

149.: Levin LA, Husberg M, Sobocinski PD, Kull VF, Friberg L, Rosenqvist M et al. A cost-effectiveness analysis of screening for silent atrial fibrillation after ischaemic stroke. Europace: European Pacing, Arrhythmias, and Cardiac Electrophysiology. 2014; 17(2):207–214 [PubMed: 25349228]

150.: Lewis M, Parker D, Weston C, Bowes M. Screening for atrial fibrillation: sensitivity and specificity of a new methodology. British Journal of General Practice. 2011; 61(582):38–39 [PMC free article: PMC3020047] [PubMed: 21401988]

151.: Li KHC, White FA, Tipoe T, Liu T, Wong MC, Jesuthasan A et al. The current state of mobile phone apps for monitoring heart rate, heart rate variability, and atrial fibrillation: Narrative review. JMIR MHealth and UHealth. 2019; 7(2):e11606 [PMC free article: PMC6396075] [PubMed: 30767904]

152.: Liao J, Khalid Z, Scallan C, Morillo C, O’Donnell M. Noninvasive cardiac monitoring for detecting paroxysmal atrial fibrillation or flutter after acute ischemic stroke: a systematic review. Stroke. 2007; 38(11):2935–2940 [PubMed: 17901394]

153.: Lin CT, Chang KC, Lin CL, Chiang CC, Lu SW, Chang SS et al. An intelligent telecardiology system using a wearable and wireless ECG to detect atrial fibrillation. IEEE Transactions on Information Technology in Biomedicine. 2010; 14(3):726–733 [PubMed: 20371411]

154.: Liu J, Fang PH, Hou Y, Li XF, Liu Y, Wang YS et al. The value of transtelephonic electrocardiogram monitoring system during the “Blanking Period” after ablation of atrial fibrillation. Journal of Electrocardiology. 2010; 43(6):667–672 [PubMed: 20667550]

155.: Lowe A, Oh TH, Stewart R. Screening for atrial fibrillation during automatic blood pressure measurements. IEEE Journal of Translational Engineering in Health and Medicine. 2018; 6:4400307 [PMC free article: PMC6204330] [PubMed: 30405979]

156.: Lown M, Yue AM, Shah BN, Corbett SJ, Lewith G, Stuart B et al. Screening for atrial fibrillation using economical and accurate technology (from the SAFETY study). American Journal of Cardiology. 2018; 122(8):1339–1344 [PubMed: 30131106]

157.: Lowres N, Mulcahy G, Gallagher R, Ben Freedman S, Marshman D, Kirkness A et al. Self-monitoring for atrial fibrillation recurrence in the discharge period post-cardiac surgery using an iPhone electrocardiogram. European Journal of Cardio-Thoracic Surgery. 2016; 50(1):44–51 [PubMed: 26850266]

158.: Lowres N, Neubeck L, Salkeld G, Krass I, McLachlan AJ, Redfern J et al. Feasibility and cost-effectiveness of stroke prevention through community screening for atrial fibrillation using iPhone ECG in pharmacies: The SEARCH-AF study. Thrombosis and Haemostasis. 2014; 111(6):1167–1176 [PubMed: 24687081]

159.: Lumikari TJ, Putaala J, Kerola A, Sibolt G, Pirinen J, Pakarinen S et al. Continuous 4-week ECG monitoring with adhesive electrodes reveals AF in patients with recent embolic stroke of undetermined source. Annals of Noninvasive Electrocardiology. 2019; 21(5):e12649 [PMC free article: PMC6850068] [PubMed: 31045315]

160.: Lyckhage LF, Hansen ML, Toft JC, Larsen SL, Brendorp B, Ali AM et al. Continuous electrocardiography for detecting atrial fibrillation beyond 1 year after stroke in primary care. Heart. 2020; 10.1136/heartjnl-2020-316904 [PMC free article: PMC8005802] [PubMed: 32620555] [CrossRef]

161.: Mant J, Fitzmaurice DA, Hobbs FD, Jowett S, Murray ET, Holder R et al. Accuracy of diagnosing atrial fibrillation on electrocardiogram by primary care practitioners and interpretative diagnostic software: analysis of data from screening for atrial fibrillation in the elderly (SAFE) trial. BMJ. 2007; 335(7616):380 [PMC free article: PMC1952490] [PubMed: 17604299]

162.: Marazzi G, Iellamo F, Volterrani M, Lombardo M, Pelliccia F, Righi D et al. Comparison of Microlife BP A200 Plus and Omron M6 blood pressure monitors to detect atrial fibrillation in hypertensive patients. Advances in Therapy. 2012; 29(1):64–70 [PubMed: 22198902]

163.: Martinek M, Aichinger J, Nesser HJ, Ziegler PD, Purerfellner H. New insights into long-term follow-up of atrial fibrillation ablation: full disclosure by an implantable pacemaker device. Journal of Cardiovascular Electrophysiology. 2007; 18(8):818–823 [PubMed: 17573835]

164.: McManus DD, Chong JW, Soni A, Saczynski JS, Esa N, Napolitano C et al. PULSE-SMART: pulse-based arrhythmia discrimination using a novel smartphone application. Journal of Cardiovascular Electrophysiology. 2016; 27(1):51–57 [PMC free article: PMC4768310] [PubMed: 26391728]

165.: McManus DD, Lee J, Maitas O, Esa N, Pidikiti R, Carlucci A et al. A novel application for the detection of an irregular pulse using an iPhone 4S in patients with atrial fibrillation. Heart Rhythm. 2013; 10(3):315–319 [PMC free article: PMC3698570] [PubMed: 23220686]

166.: Mehta DD, Nazir NT, Trohman RG, Volgman AS. Single-lead portable ECG devices: perceptions and clinical accuracy compared to conventional cardiac monitoring. Journal of Electrocardiology. 2015; 48(4):710–716 [PubMed: 25968916]

167.: Miracapillo G, Addonisio L, Breschi M, F DES, Manfredini E, Corbucci G et al. Left axillary implantation of loop recorder versus the traditional left chest area: a prospective randomized study. Pacing and Clinical Electrophysiology. 2016; 39(8):830–836 [PubMed: 27119309]

168.: Mittal S, Pokushalov E, Romanov A, Ferrara M, Arshad A, Musat D et al. Long-term ECG monitoring using an implantable loop recorder for the detection of atrial fibrillation after cavotricuspid isthmus ablation in patients with atrial flutter. Heart Rhythm. 2013; 10(11):1598–1604 [PubMed: 23911429]

169.: Montenero AS, Quayyum A, Franciosa P, Mangiameli D, Antonelli A, Barbieri L et al. Implantable loop recorders: a novel method to judge patient perception of atrial fibrillation. Preliminary results from a pilot study. Journal of Interventional Cardiac Electrophysiology. 2004; 10(3):211–220 [PubMed: 15133357]

170.: Morgan S, Mant D. Randomised trial of two approaches to screening for atrial fibrillation in UK general practice. British Journal of General Practice. 2002; 52(478):373–374, 377–380 [PMC free article: PMC1314292] [PubMed: 12014534]

171.: Mulder AA, Wijffels MC, Wever EF, Kelder JC, Boersma LV. Arrhythmia detection after atrial fibrillation ablation: value of incremental monitoring time. Pacing and Clinical Electrophysiology. 2012; 35:164–169 [PubMed: 21883308]

172.: Muller A, Scharner W, Borchardt T, Och W, Korb H. Reliability of an external loop recorder for automatic recognition and transtelephonic ECG transmission of atrial fibrillation. Journal of Telemedicine and Telecare. 2009; 15(8):391–396 [PubMed: 19948705]

173.: Narasimha D, Hanna N, Beck H, Chaskes M, Glover R, Gatewood R et al. Validation of a smartphone-based event recorder for arrhythmia detection. Pacing and Clinical Electrophysiology. 2018; 41(5):487–494 [PubMed: 29493801]

174.: National Institute for Health and Care Excellence. Developing NICE guidelines: the manual [Updated October 2018]. London. National Institute for Health and Care Excellence, 2014. Available from: https://www.nice.org.uk/process/pmg20/chapter/introduction-and-overview [PubMed: 26677490]

175.: Nault I, Andre P, Plourde B, Leclerc F, Sarrazin JF, Philippon F et al. Validation of a novel single lead ambulatory ECG monitor - CardiostatTM - Compared to a standard ECG Holter monitoring. Journal of Electrocardiology. 2019; 53:57–63 [PubMed: 30641305]

176.: Nemati S, Ghassemi MM, Ambai V, Isakadze N, Levantsevych O, Shah A et al. Monitoring and detecting atrial fibrillation using wearable technology. Conference proceedings: Annual International Conference of the IEEE Engineering in Medicine and Biology Society. 2016; 2016:3394–3397 [PubMed: 28269032]

177.: Nigolian A, Dayal N, Nigolian H, Stettler C, Burri H. Diagnostic accuracy of multi-lead ECGs obtained using a pocket-sized bipolar handheld event recorder. Journal of Electrocardiology. 2018; 51(2):278–281 [PubMed: 29223306]

178.: Nolker G, Mayer J, Boldt LH, Seidl K, V VAND, Massa T et al. Performance of an implantable cardiac monitor to detect atrial fibrillation: results of the DETECT AF study. Journal of Cardiovascular Electrophysiology. 2016; 27(12):1403–1410 [PubMed: 27565119]

179.: Omboni S, Verberk WJ. Opportunistic screening of atrial fibrillation by automatic blood pressure measurement in the community. BMJ Open. 2016; 6(4):e010745 [PMC free article: PMC4838727] [PubMed: 27072571]

180.: Oncu MR, Ozdemir F, Aydin SA. Evaluation of accuracy and reliability of electrocardiographs interpreted by emergency medicine assistants. Eastern Journal of Medicine. 2019; 24(4):427–433

181.: Orchard J, Lowres N, Freedman SB, Ladak L, Lee W, Zwar N et al. Screening for atrial fibrillation during influenza vaccinations by primary care nurses using a smartphone electrocardiograph (iECG): A feasibility study. European Journal of Preventive Cardiology. 2016; 23(2 suppl):13–20 [PubMed: 27892421]

182.: Osaka Y, Takigawa M, Takahashi A, Kuwahara T, Okubo K, Takahashi Y et al. The proportion of asymptomatic recurrence after catheter ablation of atrial fibrillation in patients with a pacemaker for sick sinus syndrome. Indian Pacing and Electrophysiology Journal. 2017; 17(5):125–131 [PMC free article: PMC5652287] [PubMed: 29192587]

183.: Osako M, Sato T, Fujii H, Sumita T, Fujiwara H, Nakao Y et al. Reliability and efficacy of a monitoring system for an implanted pulse generator. Journal of Artificial Organs. 2002; 5(4):233–238

184.: Osca Asensi J, Izquierdo de Francisco MT, Cano Perez O, Sancho Tello de Carranza MJ, Alberola Rubio J, Planells Palop C et al. The RITHMI study: diagnostic ability of a heart rhythm monitor for automatic detection of atrial fibrillation. Revista Española de Cardiología. 2020; 10.1016/j.rec.2020.05.034 [PubMed: 32792313] [CrossRef]

185.: Pagola J, Juega J, Francisco-Pascual J, Moya A, Sanchis M, Bustamante A et al. Yield of atrial fibrillation detection with Textile Wearable Holter from the acute phase of stroke: pilot study of Crypto-AF registry. International Journal of Cardiology. 2018; 251:45–50 [PubMed: 29107360]

186.: Park YM, Lee DI, Park HC, Shim J, Choi JI, Lim HE et al. Feasibility and accuracy of a new mobile electrocardiography device, ER-2000, in the diagnosis of arrhythmia. Journal of Arrhythmia. 2015; 31(4):201–209 [PMC free article: PMC4555462] [PubMed: 26336560]

187.: Pastor-Perez FJ, Manzano-Fernandez S, Goya-Esteban R, Pascual-Figal DA, Barquero-Perez O, Rojo-Alvarez JL et al. Comparison of detection of arrhythmias in patients with chronic heart failure secondary to non-ischemic versus ischemic cardiomyopathy by 1 versus 7-day holter monitoring. American Journal of Cardiology. 2010; 106(5):677–681 [PubMed: 20723645]

188.: Pedersen KB, Chemnitz A, Madsen C, Sandgaard NCF, Bak S, Brandes A. Low incidence of atrial fibrillation in patients with transient ischemic attack. Cerebrovascular Diseases Extra. 2016; 6(3):140–149 [PMC free article: PMC5216214] [PubMed: 27898406]

189.: Perez-Valero J, Victoria Caballero Pintado M, Melgarejo F, Garcia-Sanchez AJ, Garcia-Haro J, Cordoba FG et al. Symbolic recurrence analysis of RR interval to detect atrial fibrillation. Journal of Clinical Medicine. 2019; 8(11):1840

190.: Philippsen TJ, Christensen LS, Hansen MG, Dahl JS, Brandes A. Detection of subclinical atrial fibrillation in high-risk patients using an insertable cardiac monitor. JACC: Clinical Electrophysiology. 2017; 3(13):1557–1564 [PubMed: 29759838]

191.: Plummer CJ, Henderson S, Gardener L, McComb JM. The use of permanent pacemakers in the detection of cardiac arrhythmias. Europace: European Pacing, Arrhythmias, and Cardiac Electrophysiology. 2001; 3(3):229–232 [PubMed: 11467465]

192.: Plummer CJ, McComb JM, trial SA. Detection of atrial fibrillation by permanent pacemakers: observations from the STOP AF trial. Cardiac Electrophysiology Review. 2003; 7(4):333–340 [PubMed: 15071248]

193.: Podd SJ, Sugihara C, Furniss SS, Sulke N. Are implantable cardiac monitors the ‘gold standard’ for atrial fibrillation detection? A prospective randomized trial comparing atrial fibrillation monitoring using implantable cardiac monitors and DDDRP permanent pacemakers in post atrial fibrillation ablation patients. Europace: European Pacing, Arrhythmias, and Cardiac Electrophysiology. 2016; 18(7):1000–1005 [PubMed: 26585596]

194.: Poh MZ, Poh YC, Chan PH, Wong CK, Pun L, Leung WW et al. Diagnostic assessment of a deep learning system for detecting atrial fibrillation in pulse waveforms. Heart. 2018; 104(23):1921–1928 [PubMed: 29853485]

195.: Poon K, Okin PM, Kligfield P. Diagnostic performance of a computer-based ECG rhythm algorithm. Journal of Electrocardiology. 2005; 38(3):235–238 [PubMed: 16003708]

196.: Poulsen MB, Binici Z, Dominguez H, Soja AM, Kruuse C, Hornnes AH et al. Performance of short ECG recordings twice daily to detect paroxysmal atrial fibrillation in stroke and transient ischemic attack patients. International Journal of Stroke. 2017; 12(2):192–196 [PubMed: 27694312]

197.: Proesmans T, Mortelmans C, Van Haelst R, Verbrugge F, Vandervoort P, Vaes B. Mobile phone-based use of the photoplethysmography technique to detect atrial fibrillation in primary care: Diagnostic accuracy study of the FibriCheck app. JMIR MHealth and UHealth. 2019; 7(3):e12284 [PMC free article: PMC6456825] [PubMed: 30916656]

198.: Proietti M, Farcomeni A, Goethals P, Scavee C, Vijgen J, Blankoff I et al. Cost-effectiveness and screening performance of ECG handheld machine in a population screening programme: The Belgian Heart Rhythm Week screening programme. European Journal of Preventive Cardiology. 2019; 26(9):964–972 [PubMed: 30935219]

199.: Purerfellner H, Pokushalov E, Sarkar S, Koehler J, Zhou R, Urban L et al. P-wave evidence as a method for improving algorithm to detect atrial fibrillation in insertable cardiac monitors. Heart Rhythm. 2014; 11(9):1575–1583 [PubMed: 24912139]

200.: Purerfellner H, Sanders P, Sarkar S, Reisfeld E, Reiland J, Koehler J et al. Adapting detection sensitivity based on evidence of irregular sinus arrhythmia to improve atrial fibrillation detection in insertable cardiac monitors. Europace: European Pacing, Arrhythmias, and Cardiac Electrophysiology. 2018; 20(FI_3):f321–f328 [PMC free article: PMC6277148] [PubMed: 29036652]

201.: Rajakariar K, Koshy AN, Sajeev JK, Nair S, Roberts L, Teh AW. Accuracy of a smartwatch based single-lead electrocardiogram device in detection of atrial fibrillation. Heart. 2020; 106(9):665–670 [PubMed: 31911507]

202.: Rajakariar K, Koshy AN, Sajeev JK, Nair S, Roberts L, Teh AW. Modified positioning of a smartphone based single-lead electrocardiogram device improves detection of atrial flutter. Journal of Electrocardiology. 2018; 51(5):884–888 [PubMed: 30177334]

203.: Ramkumar S, Nerlekar N, D’Souza D, Pol DJ, Kalman JM, Marwick TH. Atrial fibrillation detection using single lead portable electrocardiographic monitoring: a systematic review and meta-analysis. BMJ Open. 2018; 8(9):e024178 [PMC free article: PMC6144487] [PubMed: 30224404]

204.: Reiffel JA, Schwarzberg R, Murry M. Comparison of autotriggered memory loop recorders versus standard loop recorders versus 24-hour Holter monitors for arrhythmia detection. American Journal of Cardiology. 2005; 95(9):1055–1059 [PubMed: 15842970]

205.: Reiffel JA, Verma A, Kowey PR, Halperin JL, Gersh BJ, Elkind MSV et al. Rhythm monitoring strategies in patients at high risk for atrial fibrillation and stroke: a comparative analysis from the REVEAL AF study. American Heart Journal. 2020; 219:128–136 [PubMed: 31862084]

206.: Reinsch N, Ruprecht U, Buchholz J, Diehl RR, Kalsch H, Neven K. The BioMonitor 2 insertable cardiac monitor: clinical experience with a novel implantable cardiac monitor. Journal of Electrocardiology. 2018; 51(5):751–755 [PubMed: 30177307]

207.: Rekhviashvili A, Baganashvili E, Tan KY, Raymakers F, Sakandelidze T. Reproducibility and diagnostic value of E100 event recorder for patients with complains on heart arrhythmias and no changes on multiple routine ECGs and 24-hour holter monitoring. Georgian Medical News. 2012; (203):29–33 [PubMed: 22466537]

208.: Renier W, Geelen M, Steverlynck L, Wauters J, Aertgeerts B, Verbakel J et al. Can the heartscan be used for diagnosis and monitoring of emergencies in general practice? Acta Cardiologica. 2012; 67(5):525–531 [PubMed: 23252002]

209.: Reverberi C, Rabia G, De Rosa F, Bosi D, Botti A, Benatti G. The RITMIATM smartphone app for automated detection of atrial fibrillation: accuracy in consecutive patients undergoing elective electrical cardioversion. BioMed Research International. 2019; 2019:4861951 [PMC free article: PMC6634013] [PubMed: 31355264]

210.: Rhys GC, Azhar MF, Foster A. Screening for atrial fibrillation in patients aged 65 years or over attending annual flu vaccination clinics at a single general practice. Quality in Primary Care. 2013; 21(2):131–140 [PubMed: 23735694]

211.: Ricci R, Puglisi A, Azzolini P, Spampinato A, Pignalberi C, Bellocci F et al. Reliability of a new algorithm for automatic mode switching from DDDR to DDIR pacing mode in sinus node disease patients with chronotropic incompetence and recurrent paroxysmal atrial fibrillation. Pacing and Clinical Electrophysiology. 1996; 19(11 Pt 2):1719–1723 [PubMed: 8945030]

212.: Rincon F, Grassi PR, Khaled N, Atienza D, Sciuto D. Automated real-time atrial fibrillation detection on a wearable wireless sensor platform. Conference proceedings: Annual International Conference of the IEEE Engineering in Medicine and Biology Society. 2012; 2012:2472–2475 [PubMed: 23366426]

213.: Ritter MA, Kochhauser S, Duning T, Reinke F, Pott C, Dechering DG et al. Occult atrial fibrillation in cryptogenic stroke: detection by 7-day electrocardiogram versus implantable cardiac monitors. Stroke. 2013; 44(5):1449–1452 [PubMed: 23449264]

214.: Rizos T, Rasch C, Jenetzky E, Hametner C, Kathoefer S, Reinhardt R et al. Detection of paroxysmal atrial fibrillation in acute stroke patients. Cerebrovascular Diseases. 2010; 30(4):410–417 [PubMed: 20720410]

215.: Roche F, Gaspoz JM, Da Costa A, Isaaz K, Duverney D, Pichot V et al. Frequent and prolonged asymptomatic episodes of paroxysmal atrial fibrillation revealed by automatic long-term event recorders in patients with a negative 24-hour Holter. Pacing and Clinical Electrophysiology. 2002; 25(11):1587–1593 [PubMed: 12494616]

216.: Rojo-Martinez E, Sandin-Fuentes M, Calleja-Sanz AI, Cortijo-Garcia E, Garcia-Bermejo P, Ruiz-Pinero M et al. High performance of an implantable Holter monitor in the detection of concealed paroxysmal atrial fibrillation in patients with cryptogenic stroke and a suspected embolic mechanism. Revista de Neurología. 2013; 57(6):251–257 [PubMed: 24008935]

217.: Rosenberg MA, Samuel M, Thosani A, Zimetbaum PJ. Use of a noninvasive continuous monitoring device in the management of atrial fibrillation: A pilot study. Pacing and Clinical Electrophysiology. 2013; 36(3):328–333 [PMC free article: PMC3618372] [PubMed: 23240827]

218.: Ross LS, Bettin M, Kochhauser S, Ritter M, Minnerup J, Eckardt L et al. Sensitive detection of atrial fibrillation in acute stroke patients by short-term bedside electrocardiography monitoring software analysis. Cerebrovascular Diseases. 2018; 45(1–2):54–60 [PubMed: 29402843]

219.: Roten L, Schilling M, Haberlin A, Seiler J, Schwick NG, Fuhrer J et al. Is 7-day event triggered ECG recording equivalent to 7-day Holter ECG recording for atrial fibrillation screening? Heart. 2012; 98(8):645 [PubMed: 22397942]

220.: Rozen G, Vaid J, Hosseini SM, Kaadan MI, Rafael A, Roka A et al. Diagnostic accuracy of a novel mobile phone application for the detection and monitoring of atrial fibrillation. American Journal of Cardiology. 2018; 121(10):1187–1191 [PubMed: 29525063]

221.: Ryabykina GV, Shokhzodaeva ZO, Sapelnikov OV, Makeev MI, Kozhemyakina ES, Shchedrina EV et al. Diagnostic utility of long-term remote ECG monitoring in compare with 24 hour Holter monitoring in patients with atrial fibrillation after catheter radiofrequency ablation in the early postoperative period. Terapevticheskii Arkhiv. 2018; 90(12):12–16 [PubMed: 30701827]

222.: Sabar MI, Ara F, Henderson A, Ahmed O, Potter C, John I et al. A study to assess a novel automated electrocardiogram technology in screening for atrial fibrillation. Pacing and Clinical Electrophysiology. 2019; 42:1383–1389 [PubMed: 31482579]

223.: Sack S, Mouton E, Defaye P, Dagres N, Wolfhard U, Wieneke H et al. Improved detection and analysis of sensed and paced events in dual chamber pacemakers with extended memory function. A prospective multicenter trial in 626 patients. Herz. 2001; 26(1):30–39 [PubMed: 11258107]

224.: Salvatori V, Becattini C, Laureti S, Baglioni G, Germini F, Grilli P et al. Holter monitoring to detect silent atrial fibrillation in high-risk subjects: the Perugia General Practitioner Study. Internal and Emergency Medicine. 2015; 10(5):595–601 [PubMed: 25944128]

225.: Samol A, Masin M, Gellner R, Otte B, Pavenstadt HJ, Ringelstein EB et al. Prevalence of unknown atrial fibrillation in patients with risk factors. Europace: European Pacing, Arrhythmias, and Cardiac Electrophysiology. 2013; 15(5):657–662 [PubMed: 23258819]

226.: Sanak D, Hutyra M, Kral M, Bartkova A, Zapletalova J, Fedorco M et al. Atrial fibrillation in young ischemic stroke patients: an underestimated cause? European Neurology. 2015; 73(3–4):158–163 [PubMed: 25573455]

227.: Sanak D, Hutyra M, Kral M, Bartkova A, Zapletalova J, Fedorco M et al. Paroxysmal atrial fibrillation in young cryptogenic ischemic stroke: a 3-week ECG Holter monitoring study. Biomedical Papers of the Medical Faculty of the University Palacky, Olomouc, Czechoslovakia. 2015; 159(2):283–287 [PubMed: 25916280]

228.: Sanders P, Purerfellner H, Pokushalov E, Sarkar S, Di Bacco M, Maus B et al. Performance of a new atrial fibrillation detection algorithm in a miniaturized insertable cardiac monitor: Results from the Reveal LINQ Usability Study. Heart Rhythm. 2016; 13(7):1425–1430 [PubMed: 26961298]

229.: Schaefer JR, Leussler D, Rosin L, Pittrow D, Hepp T. Improved detection of paroxysmal atrial fibrillation utilizing a software-assisted electrocardiogram approach. PloS One. 2014; 9(2):e89328 [PMC free article: PMC3938451] [PubMed: 24586692]

230.: Schuchert A, Behrens G, Meinertz T. Impact of long-term ECG recording on the detection of paroxysmal atrial fibrillation in patients after an acute ischemic stroke. Pacing and Clinical Electrophysiology. 1999; 22(7):1082–1084 [PubMed: 10456638]

231.: Schukraft S, Mancinetti M, Hayoz D, Faucherre Y, Cook S, Arroyo D et al. Handheld ECG Tracking of in-hOspital Atrial Fibrillation The HECTO-AF trial clinical study protocol. Trials [Electronic Resource]. 2019; 20(1):92 [PMC free article: PMC6354419] [PubMed: 30700332]

232.: Seidl K, Meisel E, VanAgt E, Ottenhoff F, Hess M, Hauer B et al. Is the atrial high rate episode diagnostic feature reliable in detecting paroxysmal episodes of atrial tachyarrhythmias? Pacing and Clinical Electrophysiology. 1998; 21(4 Pt 1):694–700 [PubMed: 9584299]

233.: Sejr MH, May O, Damgaard D, Sandal BF, Nielsen JC. External continuous ECG versus loop recording for atrial fibrillation detection in patients who had a stroke. Heart. 2019; 105(11):848–854 [PubMed: 30898849]

234.: Sejr MH, Nielsen JC, Damgaard D, Sandal BF, May O. Atrial fibrillation detected by external loop recording for seven days or two-day simultaneous Holter recording: A comparison in patients with ischemic stroke or transient ischemic attack. Journal of Electrocardiology. 2017; 50(3):287–293 [PubMed: 28118928]

235.: Selder JL, Breukel L, Blok S, van Rossum AC, Tulevski, II, Allaart CP. A mobile one-lead ECG device incorporated in a symptom-driven remote arrhythmia monitoring program. The first 5,982 Hartwacht ECGs. Netherlands Heart Journal. 2019; 27(1):38–45 [PMC free article: PMC6311156] [PubMed: 30523617]

236.: Shafqat S, Kelly PJ, Furie KL. Holter monitoring in the diagnosis of stroke mechanism. Internal Medicine Journal. 2004; 34(6):305–309 [PubMed: 15228390]

237.: Slocum J, Sahakian A, Swiryn S. Diagnosis of atrial fibrillation from surface electrocardiograms based on computer-detected atrial activity. Journal of Electrocardiology. 1992; 25(1):1–8 [PubMed: 1735788]

238.: Solomon MD, Yang J, Sung SH, Livingston ML, Sarlas G, Lenane JC et al. Incidence and timing of potentially high-risk arrhythmias detected through long term continuous ambulatory electrocardiographic monitoring. BMC Cardiovascular Disorders. 2016; 16:35 [PMC free article: PMC4756401] [PubMed: 26883019]

239.: Solosenko A, Petrenas A, Paliakaite B, Sornmo L, Marozas V. Detection of atrial fibrillation using a wrist-worn device. Physiological Measurement. 2019; 40(2):025003 [PubMed: 30695758]

240.: Somerville S, Somerville J, Croft P, Lewis M. Atrial fibrillation: a comparison of methods to identify cases in general practice. British Journal of General Practice. 2000; 50(458):727–729 [PMC free article: PMC1313802] [PubMed: 11050790]

241.: Sposato LA, Klein FR, Jauregui A, Ferrua M, Klin P, Zamora R et al. Newly diagnosed atrial fibrillation after acute ischemic stroke and transient ischemic attack: Importance of immediate and prolonged continuous cardiac monitoring. Journal of Stroke and Cerebrovascular Diseases. 2012; 21(3):210–216 [PubMed: 20727789]

242.: Stahrenberg R, Weber-Kruger M, Seegers J, Edelmann F, Lahno R, Haase B et al. Enhanced detection of paroxysmal atrial fibrillation by early and prolonged continuous holter monitoring in patients with cerebral ischemia presenting in sinus rhythm. Stroke. 2010; 41(12):2884–2888 [PubMed: 20966415]

243.: Stergiou GS, Karpettas N, Protogerou A, Nasothimiou EG, Kyriakidis M. Diagnostic accuracy of a home blood pressure monitor to detect atrial fibrillation. Journal of Human Hypertension. 2009; 23(10):654–658 [PubMed: 19279661]

244.: Sudlow M, Rodgers H, Kenny RA, Thomson R. Identification of patients with atrial fibrillation in general practice: a study of screening methods. BMJ. 1998; 317(7154):327–328 [PMC free article: PMC28628] [PubMed: 9685281]

245.: Suissa L, Lachaud S, Mahagne MH. Continuous ECG monitoring for tracking down atrial fibrillation after stroke: Holter or automated analysis strategy? European Neurology. 2014; 72(1–2):7–12 [PubMed: 24777038]

246.: Suissa L, Lachaud S, Mahagne MH. Optimal timing and duration of continuous electrocardiographic monitoring for detecting atrial fibrillation in stroke patients. Journal of Stroke and Cerebrovascular Diseases. 2013; 22(7):991–995 [PubMed: 22349706]

247.: Sutamnartpong P, Dharmasaroja PA, Ratanakorn D, Arunakul I. Atrial fibrillation and paroxysmal atrial fibrillation detection in patients with acute ischemic stroke. Journal of Stroke and Cerebrovascular Diseases. 2014; 23(5):1138–1141 [PubMed: 24189453]

248.: Svennberg E, Stridh M, Engdahl J, Al-Khalili F, Friberg L, Frykman V et al. Safe automatic one-lead electrocardiogram analysis in screening for atrial fibrillation. Europace: European Pacing, Arrhythmias, and Cardiac Electrophysiology. 2017; 19(9):1449–1453 [PubMed: 28339578]

249.: Swancutt D, Hobbs R, Fitzmaurice D, Mant J, Murray E, Jowett S et al. A randomised controlled trial and cost effectiveness study of systematic screening (targeted and total population screening) versus routine practice for the detection of atrial fibrillation in the over 65s: (SAFE). BMC Cardiovascular Disorders. 2004; 4:12 [PMC free article: PMC509245] [PubMed: 15283871]

250.: Swerdlow CD, Schsls W, Dijkman B, Jung W, Sheth NV, Olson WH et al. Detection of atrial fibrillation and flutter by a dual-chamber implantable cardioverter-defibrillator. For the Worldwide Jewel AF Investigators. Circulation. 2000; 101(8):878–885 [PubMed: 10694527]

251.: Taggar JS, Coleman T, Lewis S, Heneghan C, Jones M. Accuracy of methods for detecting an irregular pulse and suspected atrial fibrillation: a systematic review and meta-analysis. European Journal of Preventive Cardiology. 2016; 23(12):1330–1338 [PMC free article: PMC4952027] [PubMed: 26464292]

252.: Takagi T, Miyazaki S, Kusa S, Taniguchi H, Ichihara N, Iwasawa J et al. Role of extended external auto-triggered loop recorder monitoring for atrial fibrillation. Circulation Journal. 2014; 78(11):2637–2642 [PubMed: 25241890]

253.: Tang SC, Huang PW, Hung CS, Shan SM, Lin YH, Shieh JS et al. Identification of atrial fibrillation by quantitative analyses of fingertip photoplethysmogram. Scientific Reports. 2017; 7:45644 [PMC free article: PMC5377330] [PubMed: 28367965]

254.: Tarakji KG, Wazni OM, Callahan T, Kanj M, Hakim AH, Wolski K et al. Using a novel wireless system for monitoring patients after the atrial fibrillation ablation procedure: The iTransmit study. Heart Rhythm. 2015; 12(3):554–559 [PubMed: 25460854]

255.: Tarniceriu A, Harju J, Yousefi ZR, Vehkaoja A, Parak J, Yli-Hankala A et al. The accuracy of atrial fibrillation detection from wrist photoplethysmography. A study on post-operative patients. Conference proceedings: Annual International Conference of the IEEE Engineering in Medicine and Biology Society. 2018; 2018:1–4 [PubMed: 30440305]

256.: Tavernier R, Wolf M, Kataria V, Phlips T, Huys R, Taghji P et al. Screening for atrial fibrillation in hospitalised geriatric patients. Heart. 2018; 104(7):588–593 [PubMed: 28883032]

257.: Terranova P, Valli P, Terranova P, Dell’Orto S, Greco EM. Pacemaker prevention therapy in drug-refractory paroxysmal atrial fibrillation: reliability of diagnostics and effectiveness of prevention pacing therapy in Vitatron selection device. Indian Pacing and Electrophysiology Journal. 2006; 6(2):63–74 [PMC free article: PMC1501106] [PubMed: 16943898]

258.: Tieleman RG, Plantinga Y, Rinkes D, Bartels GL, Posma JL, Cator R et al. Validation and clinical use of a novel diagnostic device for screening of atrial fibrillation. Europace: European Pacing, Arrhythmias, and Cardiac Electrophysiology. 2014; 16(9):1291–1295 [PMC free article: PMC4149608] [PubMed: 24825766]

259.: Tison GH, Sanchez JM, Ballinger B, Singh A, Olgin JE, Pletcher MJ et al. Passive detection of atrial fibrillation using a commercially available smartwatch. JAMA Cardiology. 2018; 3(5):409–416 [PMC free article: PMC5875390] [PubMed: 29562087]

260.: Towhari J, Masud N, Alanazi H. Evaluation of the diagnostic accuracy of smartphone electrocardiogram recorder compared to standard 12 lead electrocardiography in hospital settings. Saudi Medical Journal. 2019; 40(6):575–581 [PMC free article: PMC6778753] [PubMed: 31219492]

261.: Tu HT, Chen Z, Swift C, Churilov L, Guo R, Liu X et al. Smartphone electrographic monitoring for atrial fibrillation in acute ischemic stroke and transient ischemic attack. International Journal of Stroke. 2017; 12(7):786–789 [PubMed: 28884653]

262.: Tu HT, Spence S, Kalman JM, Davis SM. Twenty-eight day Holter monitoring is poorly tolerated and insensitive for paroxysmal atrial fibrillation detection in cryptogenic stroke. Internal Medicine Journal. 2014; 44(5):505–508 [PubMed: 24816310]

263.: Turakhia MP, Hoang DD, Zimetbaum P, Miller JD, Froelicher VF, Kumar UN et al. Diagnostic utility of a novel leadless arrhythmia monitoring device. American Journal of Cardiology. 2013; 112(4):520–524 [PubMed: 23672988]

264.: Turakhia MP, Ullal AJ, Hoang DD, Than CT, Miller JD, Friday KJ et al. Feasibility of extended ambulatory electrocardiogram monitoring to identify silent atrial fibrillation in high-risk patients: the screening study for undiagnosed atrial fibrillation (STUDY-AF). Clinical Cardiology. 2015; 38(5):285–292 [PMC free article: PMC4654330] [PubMed: 25873476]

265.: Vaes B, Stalpaert S, Tavernier K, Thaels B, Lapeire D, Mullens W et al. The diagnostic accuracy of the MyDiagnostick to detect atrial fibrillation in primary care. BMC Family Practice. 2014; 15:113 [PMC free article: PMC4069340] [PubMed: 24913608]

266.: Valiaho ES, Kuoppa P, Lipponen JA, Martikainen TJ, Jantti H, Rissanen TT et al. Wrist band photoplethysmography in detection of individual pulses in atrial fibrillation and algorithm-based detection of atrial fibrillation. Europace: European Pacing, Arrhythmias, and Cardiac Electrophysiology. 2019; 21(7):1031–1038 [PubMed: 31505594]

267.: Veale EL, Stewart AJ, Mathie A, Lall SK, Rees-Roberts M, Savickas V et al. Pharmacists detecting atrial fibrillation (PDAF) in primary care during the influenza vaccination season: a multisite, cross-sectional screening protocol. BMJ Open. 2018; 8(3):e021121 [PMC free article: PMC5857694] [PubMed: 29540425]

268.: Velthuis BO, Bos J, Kraaier K, Stevenhagen J, Van Opstal JM, Van Der Palen J et al. Performance of an external transtelephonic loop recorder for automated detection of paroxysmal atrial fibrillation. Annals of Noninvasive Electrocardiology. 2013; 18(6):564–570 [PMC free article: PMC6932654] [PubMed: 24303971]

269.: Verberk WJ, De Leeuw PW. Accuracy of oscillometric blood pressure monitors for the detection of atrial fibrillation: a systematic review. Expert Review of Medical Devices. 2012; 9(6):635–640 [PubMed: 23249156]

270.: Verberk WJ, Omboni S, Kollias A, Stergiou GS. Screening for atrial fibrillation with automated blood pressure measurement: Research evidence and practice recommendations. International Journal of Cardiology. 2016; 203:465–473 [PubMed: 26547741]

271.: Vukajlovic D, Bojovic B, Hadzievski L, George S, Gussak I, Panescu D. Wireless remote monitoring of atrial fibrillation using reconstructed 12-lead ECGs. Conference proceedings: Annual International Conference of the IEEE Engineering in Medicine and Biology Society. 2010; 2010:1113–1118 [PubMed: 21096319]

272.: Wachter R. Finding atrial fibrillation in stroke - evaluation of enhanced and prolonged holter monitoring. 2013. Available from: Last accessed: 04/02/2020.

273.: Wachter R, Weber-Kruger M, Seegers J, Edelmann F, Wohlfahrt J, Wasser K et al. Age-dependent yield of screening for undetected atrial fibrillation in stroke patients: the Find-AF study. Journal of Neurology. 2013; 260(8):2042–2045 [PMC free article: PMC3734596] [PubMed: 23632947]

274.: Wang J, Wang P, Wang S. Automated detection of atrial fibrillation in ECG signals based on wavelet packet transform and correlation function of random process. Biomedical Signal Processing and Control. 2020; 55:101662

275.: Wasserlauf J, You C, Patel R, Valys A, Albert D, Passman R. Smartwatch performance for the detection and quantification of atrial fibrillation. Circulation: Arrhythmia and Electrophysiology. 2019; 12(6):e006834 [PubMed: 31113234]

276.: Welton NJ, McAleenan A, Thom HH, Davies P, Hollingworth W, Higgins JP et al. Screening strategies for atrial fibrillation: a systematic review and cost-effectiveness analysis. Health Technology Assessment. 2017; 21(29) [PubMed: 28629510]

277.: Wiegand UKH, Schneider R, Bode F, Brandes A, Diederich KW, Potratz J. Long-term atrioventricular synchrony in a single-lead VDD-pacemaker system: Comparison with a DDD-pacemaker in respect of their holter data-legends. European Journal of Cardiac Pacing and Electrophysiology. 1997; 7(4):183–191

278.: Wiesel J, Abraham S, Messineo FC. Screening for asymptomatic atrial fibrillation while monitoring the blood pressure at home: Trial of regular versus irregular pulse for prevention of stroke (TRIPPS 2.0). American Journal of Cardiology. 2013; 111(11):1598–1601 [PubMed: 23499278]

279.: Wiesel J, Arbesfeld B, Schechter D. Comparison of the microlife blood pressure monitor with the omron blood pressure monitor for detecting atrial fibrillation. American Journal of Cardiology. 2014; 114(7):1046–1048 [PubMed: 25212546]

280.: Wiesel J, Fitzig L, Herschman Y, Messineo FC. Detection of atrial fibrillation using a modified microlife blood pressure monitor. American Journal of Hypertension. 2009; 22(8):848–852 [PubMed: 19478793]

281.: Wiesel J, Wiesel D, Suri R, Messineo FC. The use of a modified sphygmomanometer to detect atrial fibrillation in outpatients. Pacing and Clinical Electrophysiology. 2004; 27(5):639–643 [PubMed: 15125721]

282.: Wiesel J, Wiesel DJ, Messineo FC. Home monitoring with a modified automatic sphygmomanometer to detect recurrent atrial fibrillation. Journal of Stroke and Cerebrovascular Diseases. 2007; 16(1):8–13 [PubMed: 17689385]

283.: William AD, Kanbour M, Callahan T, Bhargava M, Varma N, Rickard J et al. Assessing the accuracy of an automated atrial fibrillation detection algorithm using smartphone technology: The iREAD Study. Heart Rhythm. 2018; 15(10):1561–1565 [PubMed: 30143448]

284.: Williams J, Pearce K, Benett I. The effectiveness of a mobile ECG device in identifying AF: sensitivity, specificity and predictive value. British Journal of Cardiology. 2015; 22(2):70–72

285.: Willits I, Keltie K, Craig J, Sims A. WatchBP Home A for opportunistically detecting atrial fibrillation during diagnosis and monitoring of hypertension: a NICE medical technology guidance. Applied Health Economics and Health Policy. 2014; 12(3):255–265 [PMC free article: PMC4026667] [PubMed: 24664995]

286.: Winkler S, Axmann C, Schannor B, Kim S, Leuthold T, Scherf M et al. Diagnostic accuracy of a new detection algorithm for atrial fibrillation in cardiac telemonitoring with portable electrocardiogram devices. Journal of Electrocardiology. 2011; 44(4):460–464 [PubMed: 21419421]

287.: Wong KC, Klimis H, Lowres N, von Huben A, Marschner S, Chow CK. Diagnostic accuracy of handheld electrocardiogram devices in detecting atrial fibrillation in adults in community versus hospital settings: a systematic review and meta-analysis. Heart. 2020; 106(16):1211–1217 [PubMed: 32393588]

288.: Yan BP, Lai WHS, Chan CKY, Chan SC, Chan LH, Lam KM et al. Contact-free screening of atrial fibrillation by a smartphone using facial pulsatile photoplethysmographic signals. Journal of the American Heart Association. 2018; 7(8):05 [PMC free article: PMC6015414] [PubMed: 29622592]

289.: Yang P, Pu L, Yang L, Li F, Luo Z, Guo T et al. Value of implantable loop recorders in monitoring efficacy of radiofrequency catheter ablation in atrial fibrillation. Medical Science Monitor. 2016; 12(22):2846–2851 [PMC free article: PMC4993216] [PubMed: 27518153]

290.: Yang X, Li S, Zhao X, Liu L, Jiang Y, Li Z et al. Atrial fibrillation is not uncommon among patients with ischemic stroke and transient ischemic stroke in China. BMC Neurology. 2017; 17(1):207 [PMC free article: PMC5715624] [PubMed: 29202727]

291.: Yenikomshian M, Jarvis J, Patton C, Yee C, Mortimer R, Birnbaum H et al. Cardiac arrhythmia detection outcomes among patients monitored with the Zio patch system: a systematic literature review. Current Medical Research and Opinion. 2019; 35(10):1659–1670 [PubMed: 31045463]

292.: Zaprutko T, Zaprutko J, Baszko A, Sawicka D, Szalek A, Dymecka M et al. Feasibility of atrial fibrillation screening with mobile health technologies at pharmacies. Journal of Cardiovascular Pharmacology and Therapeutics. 2019; 25(2):142–151 [PubMed: 31578088]

293.: Ziegler PD, Koehler JL, Mehra R. Comparison of continuous versus intermittent monitoring of atrial arrhythmias. Heart Rhythm. 2006; 3(12):1445–1452 [PubMed: 17161787]

294.: Ziegler PD, Rogers JD, Ferreira SW, Nichols AJ, Richards M, Koehler JL et al. Long-term detection of atrial fibrillation with insertable cardiac monitors in a real-world cryptogenic stroke population. International Journal of Cardiology. 2017; 244:175–179 [PubMed: 28624331]

295.: Zwart LA, Jansen RW, Ruiter JH, Germans T, Simsek S, Hemels ME. Opportunistic screening for atrial fibrillation with a single lead device in geriatric patients. Journal of Geriatric Cardiology. 2020; 17(3):149–154 [PMC free article: PMC7118016] [PubMed: 32280331]

Appendices

Appendix A. Review protocols

Table 15. Review protocol: Diagnostic accuracy of point of care devices (PDF, 210K)

Table 16. Health economic review protocol (PDF, 206K)

Appendix B. Literature search strategies

This literature search strategy was used for the following reviews:

What are the most accurate methods for detecting atrial fibrillation in people with cardiovascular risk factors for AF and/or symptoms suggestive of AF?

The literature searches for this review are detailed below and complied with the methodology outlined in Developing NICE guidelines: the manual.¹⁷⁴

For more information, please see the Methods Report published as part of the accompanying documents for this guideline.

B.1. Clinical search literature search strategy (PDF, 323K)

B.2. Health Economics literature search strategy (PDF, 304K)

Appendix G. Health economic evidence tables

Please see evidence review A.

Appendix H. QUADAS2 risk of bias assessment

Download PDF (394K)

Appendix I. Excluded studies

I.1. Excluded clinical studies

Download PDF (267K)

I.2. Excluded health economic studies

None.

Appendix J. Research recommendations

J.1. Detection of persistent AF (PDF, 187K)

Final

Diagnostic evidence review

Developed by the National Guideline Centre, Royal College of Physicians

Disclaimer: The recommendations in this guideline represent the view of NICE, arrived at after careful consideration of the evidence available. When exercising their judgement, professionals are expected to take this guideline fully into account, alongside the individual needs, preferences and values of their patients or service users. The recommendations in this guideline are not mandatory and the guideline does not override the responsibility of healthcare professionals to make decisions appropriate to the circumstances of the individual patient, in consultation with the patient and, where appropriate, their carer or guardian.

Local commissioners and providers have a responsibility to enable the guideline to be applied when individual health professionals and their patients or service users wish to use it. They should do so in the context of local and national priorities for funding and developing services, and in light of their duties to have due regard to the need to eliminate unlawful discrimination, to advance equality of opportunity and to reduce health inequalities. Nothing in this guideline should be interpreted in a way that would be inconsistent with compliance with those duties.

NICE guidelines cover health and care in England. Decisions on how they apply in other UK countries are made by ministers in the Welsh Government, Scottish Government, and Northern Ireland Executive. All NICE guidance is subject to regular review and may be updated or withdrawn.

Bookshelf ID: NBK571339PMID: 34165927

Table 1PICO characteristics of review question

Population	People aged over 18 with symptoms suggestive of AF (including breathlessness, palpitations, syncope/dizziness, and chest discomfort) and/or with cardiovascular risk factors for AF (including TIA, stroke, Heart Failure, hypertension, valve disease). Departures from this population are allowed, but the evidence will be downgraded for indirectness.
Target condition	Atrial fibrillation
Index test(s)	Any point of care tests used to detect AF For example (non-exhaustive list): Manual pulse checking Pulse oximeters US devices Blood pressure monitors Microlife BPM Watch BP Home A Non-portable (but non-12 lead) ECG devices Portable ECG devices My Diagnostick AliveCor Kardia Smart portable devices e.g., phones, watches 12 lead ECG (when gold standard is long-term loop recording – see section below) Where the same test is used with a differing number of recordings across studies, these should be regarded as separate test strategies, and should thus be dealt with separately. Tests using differing periods of recording will also be dealt with separately. For example, pulse oximeters for 2 minutes will be in a separate category of index test to pulse oximeters used for 1 hour, and they could be compared to each other as separate index tests.
Reference standard(s)	The reference standard that is used will determine the type of AF that the measured accuracy relates to. The analyses will therefore be stratified by the reference standards used, as follows: 12 lead ECG, adjudicated by an expert clinician (usually cardiologist). This will theoretically pick up all constant AF but only a small proportion of intermittent AF cases. It is therefore really only useful for determining how well an index test can pick up constant AF. Ambulatory monitoring for >24 hrs [any device that gives a long term recording]. These should pick up all forms of AF. It is therefore a useful way gold standard for determining how well a test can pick up any AF. The ability of the tests to pick up AF vs no AF is being evaluated in this review, not the ability to differentiate between persistent and paroxysmal.
Outcomes	Diagnostic accuracy – sensitivity and specificity
Study design	Cross-sectional observational

Table 2Summary of studies included in the evidence review for detection of atrial fibrillation

Study	Population	Index test(s)	Reference standard
Antonicelli, 2012⁶	107 patients from Italy. Age 66; 57 men/50 women; Inclusion: Patients enrolled from the pre-surgical evaluation unit in the outpatient day surgery service at the National Research centre in Ancona Exclusion: None reported	3-lead tele-ECG	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Brito, 2018²³	127 patients from Switzerland. Age 62; males 64.6%; MI 22.8%; CABG 6.3%; CorAngio 33.9%; valvular Sx 7.9%; sinus at baseline 85% Inclusion: Consecutive patients admitted to the cardiology ward of Geneva University Hospital for coronarography 17.3%, electrophysiology procedure 26%, pacemaker implantation 3.9%, cardiac failure 3.9%, other 52%. Exclusion; Patients with pacemaker or cardioverter defibrillator	Beurer ME90 device – a handheld ECG recorder	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Bumgarner, 2018²⁶	100 patients from USA. Age 68.2; female 17%; warfarin 32%; DOACs 68%; CV performed 85% Inclusion: Consecutive patients with a diagnosis of AF who presented for scheduled elective CV with or without a planned transesophageal echo-cardiogram were screened for enrolment. Inclusion criteria included all adult patients age 18 to 90 years who were able to provide informed consent and willing to wear the KB before and after cardioversion Exclusion: Implanted pacemaker; defibrillator	the Kardia Band (KB) (AliveCor)	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Caldwell, 2012²⁹	157 patients from UK. Details not reported Inclusion: Consecutive patients with chronic AF attending the anticoagulation clinic, and consecutive patients with no prior diagnosis of AF attending for a routine ECG Exclusion: None reported	6 lead ECG from conventionally positioned limb electrodes (4 limb-leads) Supine 4-electrode 6-lead frontal plane ECG recording in supine using the prototype recorder placed on the lower thorax/abdomen Seated 4-electrode 6-lead frontal plane ECG prototype recording with loosened clothing only	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Chen, 2020³⁶	401 inpatients and outpatients from China; 197 female, 204 male; AF/no AF: age 70.4/59.3; hypertension 47.3%/42.2%; CHD 17.3%/26.3% Inclusion: >18 years; stable heart rhythm at time of study Exclusion: Situations where wristband could not be used such as bilateral UL disabilities, wrist colour ‘abnormalities’, severe occlusive disease, or significant UL oedema; implanted pulse generator	Amazfit Health band	12 lead ECG with interpretation by cardiologist/electrophysiologist
Cunha, 2019⁴⁹	101 patients from Portugal attending an outpatient cardiology unit. Inclusion: Aged >40 Exclusion: Previous diagnosis of atrial fibrillation being medicated with OACs; inability to communicate with the researcher; pacemakers; recent bypass; Wolff-Parkinson-White syndrome	AliveCor	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Desteghe, 2017⁵⁸	344 patients from Belgium admitted to cardiac wards in a tertiary hospital in Belgium. Patients with an implanted device comprised 17.2% of the cardiology population: 60% was actively paced, 7.3% was intermittently paced, and 32.7% was not being paced during the recordings. Based on chart review, 35.6% of the screened study population was known with AF. At the moment of the study, 11.9% showed AF on their 12-lead ECG. Of the entire AF population, the majority had paroxysmal AF (54.4%) while those in AF at the time of screening were mostly permanently in AF. Inclusion: Patients admitted to cardiac wards in a tertiary hospital in Belgium; able to give informed consent Exclusion: Age <18 years, patients in isolation, and those who were unable to hold both devices properly.	My Diagnostik AliveCor	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Diamantino, 2020⁵⁹	334 patients recruited from a primary care cardiovascular screening clinic in Brazil. Demographic data ata only available for sample that were +ve on AFSD: Age 61.8, female 50.5%, hypertension 72.2%, HF 40.2%, CAD 22.7%, major HD on standard echo 76.1% Inclusion: Unclear, but would need to come from an area of Brazil conducting cardiovascular screening Exclusion: Unclear	Atrial Fibrillation Screening Device	12 lead ECG with interpretation by cardiologist/electrophysiologist
Doliwa, 2009⁶³	100 patients from Sweden. Inclusion: Patients with atrial fibrillation, atrial flutter or sinus rhythm recruited from cardiology department. Exclusion: None reported	Thumb ECG device - Zenecor ECG	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Fallet, 2019⁷⁶	17 patients from Switzerland. Age 57 years; 12/17 mean; referred for catheter ablation of cardiac arrhythmia (not all with AF) Inclusion: Patients undergoing catheter ablation of various arrhythmias Exclusion: Not reported	Wrist-type photoplethysmographic (PPG) device	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Fan, 2019⁷⁷	112 patients from China. Mean age 58; female 46%; BMI 24.44; HF 4%; hypertension 52%; DM 27%; stroke/TIA/SE 7%; CAD 45%; vascular disease 55%; COPD 2%; renal dysfunction 4%; hepatic dysfunction 0%; sleep apnea 4%; hyperthyroidism 2%; current smoking 29%; median CHADSVASC 2; median HAS-BLED 1; OAC 18%; antiplatelets 27%; Inclusion: Aged 18 or over Exclusion: Patients unable to use mobile phones and smart bands, with mental or memory problems, or with a pacemaker or implantable cardioverter defibrillator.	Huawei mate 9 mobile phone – PPG measurements Huawei Honor 7x mobile phone – PPG measurements Smart band – Huawei band 2	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Gandolfo, 2015⁷⁹	207 stroke unit inpatients from Italy; 103 women; mean age 77.7 years; 86.5% recent ischaemic CVA/TIA; 13.5% haemorrhagic stroke; within 48 hour window post stroke Inclusion: Patients admitted to stroke unit because of recent (<48 hours) TIA/stroke Exclusion: Patients with rhythm controlled by pacemakers or defibrillators	Microlife AFib BP device	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Greg, 2008⁸²	1785 patients from USA (1 ECG per patient). Male 1090/1785; age 62 (male) and 63 (female); 109/1785 with AF on gold standard 12 lead testing; no other information given, apart from the fact that the 1785 ECGs had been taken from a random selection of 50000 ECGs collected from 2 teaching hospitals Inclusion: Not reported Exclusion: ECGs with extreme artefact and paced rhythm	Using the Philips resting 12-lead ECG algorithm, the index tests were Computer interpretation of full 12 lead ECG V1-V6 Computer interpretation of V2, V5 leads information only Computer interpretation of V1, V4 leads information only	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Guan, 2020⁸⁶	1479 people randomly selected from the community, in Jiangsu, China. Male: 51.7%; hypertension: 86.9%; DM: 27.4%; history of stroke: 19.3% Inclusion: Aged >50 years Exclusion: Tremors; unable to use index device properly	Snap ECG – portable single lead (blinded). 1 minute reading	12 lead supine ECG (10s) read by 1 cardiologist
Haberman, 2015⁹⁰	130 patients from USA (there were 251 other participants form other populations also analysed, such as athletes and asymptomatic students, but the 130 are the cardiology clinic patients of relevance to this review) Age 59; male 56%; mean HR 72 Inclusion: Ambulatory cardiology patients Exclusion: Not reported	AliveCor device	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Hald, 2017⁹¹	87 from Denmark who had irregular pulse on palpation, who were also given ECG by GP/nurse (index test) and ECG by cardiologist (gold standard). The entire study looked at 970 people who were all given pulse palpation. However the larger group of 970 are not considered here because the only people given the gold standard (ECG interpreted by AF specialist) were the 87 with the irregular pulse. Hence the accuracy of pulse palpation is not determinable as we have no gold standard data on those who were negative on pulse palpation. Data not available for subset who had irregular pulse; however for our subset all had irregular pulse on palpation which makes them have a high prevalence of AF (11%) Inclusion: Any person aged >=65 from the GP practices; no previous AF; presentation was for a genuine medical reason and not for the screening itself; also positive palpation findings, but that is only for the diagnostic accuracy analysis pertinent to this review. Exclusion: Not reported	12 lead ECG carried out and interpreted by GP/nurse	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Haverkamp, 2019⁹⁶	94 patients from Norway.37% female; mean age 58; Inclusion: People having ongoing scECG cardiac surveillance who were admitted to the cardiac ward at a university hospital. Exclusion: None reported	ECG check	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Himmelreich, 2019¹⁰¹	219 people from Holland. Mean age 64.1; 53.7% male; hypertension 40.7%; DM 30.8%; hypercholesterolaemia 25.2%; known AF or AFL 10.7%; CHD 9.8%; TIA/stroke 6.1%; HF 3.7%; PVD 8.9%; CRF 12.1%; indication for inclusion: 44.4% palpitations, 43.5% other chest symptoms, 21.3% dyspnea, 14.8% lightheadedness 14.8%; fatigue 13%, collapse 2.8%, other 15.7% Eligible patients were aged 18 years or older who were assigned to 12L-ECG for any non-acute indication as ordered by the local primary care physician in 1 of 10 participating general practices across the Netherlands. Exclusion criteria were a clinically acute indication for ECG as defined by the local primary care physician (eg, suspicion of acute coronary syndrome) and presence of a pacemaker rhythm on 12L-ECG.	KardiaMobile (AliveCor, Inc) - smartphone-connected, 1L-ECG device that displays ECG recordings in real time (30 seconds) via a smartphone application with a built-in AF detection algorithm.	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Kaleschke, 2009¹¹⁷	508 patients from Germany. 66% male; mean age 61.4; mean BMI 26.6; Inclusion: Clinical indication for 12 lead surface ECG; No other details provided. Exclusion: <18 years; pacemaker or defibrillator	Omron Heartscan	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Kao, 2018¹²³	63 (1 excluded as not fulfilling inclusion criteria) people from Taiwan, recruited from emergency department; age 67; 56% male; AF 29/62 Inclusion: Aged >20 years; either with AF or no AF (diagnosed by 12 lead ECG). Exclusion: People exposed to high frequency surgical equipment during testing’ people with cardiac pacemakers or implantable defibrillators; pregnant women	Heart Spectrum Blood Pressure Monitor	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Kearley, 2014¹²⁸	1000 patients from UK. mean age 79.7; 49.3% male; Hx of AF 11%; HF 3.1; hypertension 53%; DM 12.2%; Stroke 3.1%; TIA 6.5%; Patients with AF on AADs 8.7% Inclusion: Participants aged 75 or over, living at home from 6 General practices in the UK Exclusion: People with pacemakers and defibrillators; unable to give consent; terminal illness; other reasons why participation is inappropriate at discretion of GP;	Watch BP Omron HCG-801 Merlin ECG event recorder	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Koltowski, 2019¹³³	100 patients from Poland. Mean age 68; male 66%; patients at a tertiary cardiovascular care center, admitted for hospital elective and treatment procedures for various cardiac conditions.; body mass 80.7kg; BMI 28; smoking history 43.5%; DM 20.4%; hypertension 68.4%; dyslipidemia 46.4%; CKD 32.7%; thyroid dysfunction 18.4%; COPD 6.12%; Stroke 17.35%; PAD 12.24%; stable angina 47.4%; ACS 15.31%; MI 25.5%; PCI/CABG 27.6%; other cardiac surgery 3.1%; HF 43.9%; LVEF 49%; AF 34.7%; CIED implanted 34.7%; pacemaker 24.5%; ablation 6.1% Inclusion: Undergoing regular 12-lead ECG due to standard diagnosis on admission in stable state Exclusion: Need for urgent medical care	Kardia mobile ECG (Lead I)	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Kristensen, 2016¹³⁸	93 patients from Denmark. 54% male; age 67; IHD 11%; hypertension 54%; DM 21%; known AF diagnosis 36%; Medication affecting heart rhythm 47% Inclusion: Patients from a GP clinic in Aalborg, Denmark, who performed a routine 12-lead ECG were invited to participate. The invited patients either had known paroxysmal AF or were invited among patients who came for an annual routine health check. The aim was to include 30–50% with a diagnosis of AF and 50–70% without AF. Thus this was not a consecutive sample. Exclusion: Patients with severe dementia, mental illness or poor ECG readings	Portable ECG monitor (3 lead)	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Kvist, 2019¹⁴⁰	1340 people from Denmark. Age 69; 100% male; BMI 27.3; self-reported AF 7.9%; DM 10.9%; Hypertension 42.4%; Ischaemic stroke 6.1%; acute MI 6.2%; PAD 2.2%; CABG or PCI 8.3%; COPD 6.8%; never smoked 33.9%; OACs 8.5%; AADs 1.1%; statins 35.6% Inclusion: Men aged 65-74 in Denmark Exclusion; None applied	12-lead ECG recorded (Schiller Cardiovit AT-102, Schiller Cardiovit AT-102 Plus or Philips PageWriter Trim II). The 12-lead ECGs were examined for AF by one of four study nurses	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Langley, 2012¹⁴⁵	2124 patients on a validation database from Tanzania. There was also a derivation database comprising 167 patients from UK, but these were used to derive the thresholds of algorithms and not pertinent to this review. Inclusion; Aged >70; residing in Hai district of Northern Tanzania; Exclusion: None reported	12 lead ECG, using the following automated detection algorithms, each based on a short 10s recording, were tested: Based on a co-efficient of variation of the beat intervals (CV). Threshold set at 0.12 Based on the mean successive beat interval difference (defined as the mean absolute successive beat interval difference divided by the mean beat interval (Delta). Threshold set at 0.11 Based on the co-efficient of sample entropy (COSEn). Threshold set at −1.19	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Lewis, 2011¹⁵⁰	594 patients from UK and USA. Aged >60 years; not specifically patients with cardiac symptoms or diagnoses. Inclusion and exclusion criteria not reported.	Finger-probe instrument (as used in pulse oximetry) that utilises the principle of photoplethysmography	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Lin, 2010¹⁵³	20 people from Taiwan with AF (each with 60 × 6 second tests, each counting as a single test). Therefore 1200 data points (person-tests). Also 10 people with no AF (each with 20 × 15 sec tests, each counting as a single test). Therefore 200 data points (person-tests) AF patients: Age 71.4 (range 50-89 years); AF based on 12 lead ECG Non-AF: Age 71.6 years (range 57-88 years); No AF based on 12 lead ECG Inclusion and exclusion criteria not reported.	Wearable and wireless 3-lead ECG device (Medi-Trace 200, Kendall)	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Lown, 2018¹⁵⁶	418 individuals from 3 general practices in UK aged >65 both with and without a coded diagnosis of AF in their medical records were invited to attend a single screening visit at their local general practice. Mean age 73.9; 79 found to have AF Inclusion: Aged>=65; from the 3 designated general practices Exclusion: a pacemaker, were deemed unsuitable by their named General Practitioner (GP) (e.g., terminally ill and bedridden), lacked capacity, or had a previous moderate or severe skin reaction to electrode gel.	Watch BP Alive Cor PH7 FirstBeat Bodyguard	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Mant, 2007¹⁶¹ and Hobbs, 2005¹⁰⁴	A random sample of 9866 people from UK aged 65 or over was taken. A random half of these were invited for an ECG, and the remaining half were invited if opportunistic screening had previously identified them as having an irregular pulse. This led to 2595 12 lead ECGs being recorded, including 238 from opportunistic screening in 2001-3. Inclusion: Patients taken from 25 General practices in central England. 1 GP and 1 practice nurse involved in the study. All practitioners had 1 hour training on AF detection. Exclusion: None reported	12 lead interpretive software 12 lead interpreted by GP Limb lead ECG interpreted by GP Chest lead ECG interpreted by GP 12 lead interpreted by practice nurse Limb lead ECG interpreted by practice nurse Chest lead ECG interpreted by practice nurse 12 lead interpretive software combined with GP interpretation (positive if either or both is positive)	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Marazzi, 2012¹⁶²	550 patients from Italy. Mean age 67 years; 54.3% male; bp 139.8/86.9 Inclusion: Patients referred to hypertension clinic Exclusion: <18 years; pacemaker; implanted defibrillator	Microlife BP A200 Plus Omron M6 oscillometric device	12 lead ECG, with interpretation by cardiologist/electrophysiologist
McManus, 2013¹⁶⁵	76 (undergoing cardioversion for AF; those in AF on 12 lead ECG at pre-CV, and those in sinus rhythm on 12 lead ECG at post-CV measured with iphone device) patients from USA. Age 65.3; male 77%; white 96%; hypertension 71%; hyperlipidaemia 62%; current smoking 8%; DM 28%; CAD 29%; CHF 21%; sleep apnea 16%; 11% CABG; prior cardioversion 27%; stroke 12% Inclusion: Patients with persistent AF on a roster of patients scheduled to have elective cardioversion for AF Exclusion: Not reported	iPhone 4S camera PPG measures on fingertip	12 lead ECG, with interpretation by cardiologist/electrophysiologist
McManus, 2016¹⁶⁴	128 people from USA. Age 66.2yrs; non-white 7%; 18% women; hypertension 75.7%; DM 28.2%; CAD 25%; CHF 32.8%; stroke 13.3% Inclusion: The original PULSESMART cohort included 76 participants with AF scheduled to undergo elective cardioversion at the University of Massachusetts Medical Center (UMMC). For the present study, the sample were enriched with an additional 55 participants (22 adults with AF, 15 with PACs, and 15 with PVCs) to create a cohort comprised of a more representative array of benign (PAC and PVC) and malignant (AF) causes of an irregular pulse. Patients with frequent PACs or PVCs were identified from a roster of inpatients on the cardiac telemetry unit at the UMMC. Study staff performed a review of hospital telemetry recordings on a daily basis to identify patients with frequent ectopy. Exclusion; Not reported	PPG measures on an iphone 4S	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Nigolian, 2018¹⁷⁷	52 people from Switzerland. Age 69; male 58%; pacemaker 10%; hypertension 60%; DM 21%; COPD 8%; AF on 12 lead ECG 31%; OACs 40% Inclusion: Consecutive patients admitted to the cardiology department at a University Hospital Exclusion; <18 years; inability or unwilling to consent	Beurer ME 80 device – a pocket sized (reconstructing 9 lead) ECG device	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Osca Asensi, 2020¹⁸⁴	167 patients recruited from a Cardiology outpatient clinic in Spain. SR/SF: age 54/67; hypertension 39%/54%; DM 10%/19%; OACs 54%/85% Inclusion: Patients aged >18 referred to a cardiology department for cardioversion for AF or for a general consultation (SR or AF) Exclusion: Atrial flutter or implanted pacemaker	Rithmi heart rhythm monitor: wrist monitor using PPG and ECG lead. One repetition for 3 minutes in seating	12 lead ECG read by 2 expert cardiologists
Park, 2015¹⁸⁶	17 patients from South Korea with palpitations. No other details given. Inclusion: Patients with palpitations Exclusion: None reported	mobile ECG device ER-2000s	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Poon, 2005¹⁹⁵	4297 ECGs had been taken from inpatients and outpatients in UK over a 3 week period	12 lead ECG interpreted by computer-based rhythm diagnosis (GE Healthcare Technologies MUSE software 005C, version 19)	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Proesmans, 2019¹⁹⁷	223 patients from Belgium. Age 77; male 46.6%; median (IQR) CHADSVASC 4(3-6); CHF 28.7%; DM 20.2%; stroke or TIA 22.4%; OACs 55.6%; mobile phone ownership 16.1%. From 17 GP centres. Inclusion: Known paroxysmal or persistent AF; aged >=65; other subjects without a history of AF. Exclusion: Active pacemakers	Fibricheck app (PPG) Single lead ECG using ECG-Bone	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Rajakariar, 2020²⁰¹	200 patients recruited from tertiary university hospitals in Australia. No AF/AF: age 64/76; male 64%/52%; IHD 32%/50%; hypertension 51%/50%; HF 13%/44%; DM 20%/25%; stroke/TIA 16%/17%; known AF 9%/95% Inclusion: Patients≥18 years of age admitted to the medical, cardiac or intensive care ward Exclusion: Patients with cardiac implantable electronic devices, those unable to independently use the device, or in contact isolation	Alive-Cor KardiaBand	12 lead ECG with interpretation by cardiologist/electrophysiologist
Renier, 2012²⁰⁸	177 patients from Belgium aged 55 years; 45% men Inclusion: All consecutive patients visiting ED of University hospital in Belgium; any patients hospitalised in one respiratory, one gynaecological and one orthopaedic hospital ward on one day. Exclusion: <18 years; unable to use right hand for heartscan device; did not understand language used by HCPs; no consent	Heartscan hand-held device	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Reverberi, 2019²⁰⁹	100 unselected ambulatory patients from Italy diagnosed with AF undergoing DC cardioversion; mean age 66.2; 21% female; CHADSVASC 2.3; successful CV 87.4% Inclusion: Age >18; AF undergoing CV; CHADSVASC >=2; Exclusion: Pacemaker/automatic internal cardioverter defibrillator	RITMIA HR monitor using Bluetooth to communicate with iphone app. 10 minutes.	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Rhys, 2013²¹⁰	68 patients from UK with abnormal pulses, from a screening study of 573 people, who were not already diagnosed with AF. The 68 patients with abnormal pulses were all invited to ECG but only 39 attended. No inclusion or exclusion criteria reported.	12 lead ECG interpreted by algorithm in Cardioview interpretive software (not described) 12 lead ECG interpreted by GP specialty trainee (interpretation done before sent to gold standard interpretation, so effectively blinded to gold standard)	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Rozen, 2018²²⁰	99 patients from USA (but each patient contributed two sets of data – pre-cardioversion and post-cardioversion). Patients with paroxysmal AF were referred for Holter monitoring for arrhythmia detection. 73 men/24 women; age 67.7; 91.8% white; 1% Hispanic/Latino; 1% Black; 1% Asian Inclusion: Consecutive patients with a diagnosis of AF who were scheduled for elective direct current cardioversion. Exclusion: <18 years	Cardio Rhythm Mobile Application (CRMA) (PPG)	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Sabar, 2019²²²	752 patients from UK attending a cardiology outpatient department for a routine 12 lead ECG or outpatient appointment. Age range 18-97; 51% female; no other information provided Inclusion: Age >=18; any patient attending the cardiology department for a routine 12 lead ECG or for an outpatient department Exclusion: Allergies to Velcro or metal used in device; medical condition affecting the wrists that may be interfered with by the attachment of the RhythmPad, such as a fracture necessitating a cast; pacemakers or implantable cardiac devices	6 lead ECG using Rhythm Pad device (1 × 10s). The Rhythm Pad device (Cardiocity, Lancaster, UK) is a CE-marked medical device that consists of electric potential titanium-based sensors which are placed around both arms of the patient and the right leg, using Velcro straps. The system is attached via leads to a hardware device consisting of a tablet computer that displays and stores the six-lead ECG data. An automated diagnostic report is generated at the same time, using a bespoke algorithm to determine heart rhythm and rate.	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Slocum, 1992²³⁷	82 patients from USA (for validation study, which is the relevant part for this review; the developmental study to develop the algorithm involved 73 different rhythm traces).	Algorithm for reading 12 lead ECGs. This first tested for the presence of noncoupled P waves. If noncoupled P waves were detected the rhythm was considered nonatrial fibrillation and no further testing was done. If the rhythm did not have noncoupled P waves, and the percent power in each lead II or V1 was >=32% the rhythm was considered AF. This algorithm was derived from the ‘training set’ of 72 rhythms in the developmental analysis.	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Somerville, 2000²⁴⁰	86 patients from UK. 30% with AF; no other details provided Inclusion: The study patients were all recruited from a single practice. Patients aged 65 years or over with a diagnosis of atrial fibrillation were identified by searching computerised records using the Read Codes for atrial fibrillation and digoxin prescription. An equal number of patients aged 65 years or over, without either code in their computer records, was sampled. All patients were invited to attend the surgery by appointment. Exclusion: None reported	Pulse palpation Bipolar ECGs 12 lead ECG by non-expert interpreters	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Stergiou, 2009²⁴³	73 patients from Greece. Age 70.5; 65.8% male; BMI 27; smokers 5.5%; CVD 39.7%; DM 15.1%; hypertension 63%; systolic bp 138; diastolic bp 80; AF 37% Inclusion: Subjects with known sustained AF, or other non-AF arrhythmias, and controls with sinus rhythm were recruited among those attending an Outpatients Hypertension Clinic, patients admitted in a University Department of Medicine wards and healthy volunteers. Exclusion: age <35 years, presence of a pacemaker, and/or an implanted defibrillator and refusal to participate.	Microlife BPA100 Plus	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Tieleman, 2014²⁵⁸	868 people from Holland. Inclusion: Patients visiting the outpatient cardiology clinic, or patients attending 2 GP clinics for influenza vaccination Exclusion: None reported	My Diagnostik	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Vaes, 2014²⁶⁵	191 patients from Belgium. Age 74.2; male 52.4%; BMI 26.6; CHADSVASC 3; DM 21.5%; hypertension 81.7%; CAD 13.1%; TIA/CVA 11%; PAD 4.2%; AF 53.9%; warfarin 51.8%; DOACs 20.9%; antiplatelets 15.7% Inclusion: Participating general practitioners were asked to invite patients with known, paroxysmal or chronic atrial fibrillation to participate in the study. Furthermore, this convenience sample was added up with subjects without a history of atrial fibrillation. Exclusion: Pacemaker in active mode	My Diagnostik	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Vukajlovic, 2010²⁷¹	18 (but measured pre and post CV so 36 data points) people from Serbia. Age 33-77; 12 male; Inclusion: People with AF undergoing electrical DC cardioversion Exclusion: None reported	Cardiobip, a portable handheld system for remote monitoring of patients	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Wiesel, 2004²⁸¹	450 people from USA contributing to 464 office visits (14 attended twice) 59% men; mean age 69; most common associated medical conditions were hypertension, CAD and DM Inclusion: Unselected outpatients followed by an urban cardiology practice who had an ECG performed during scheduled office visits. Exclusion: None reported	Omron 712C sphygmomanometer	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Wiesel, 2009²⁸⁰	405 patients from USA. Mean age 73; male 51%; white 82%; black 8%; other 10%; CHF 6.7%; Hypertension 51.6%; DM 14.8%; CAD 37.3% Inclusion: Unselected general cardiology outpatients seen for scheduled visits in 2 cardiology centres in NY Exclusion: Pacemakers; defibrillators	Microlife BP3MQ1-2D BP monitor	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Wiesel, 2014²⁷⁹	183 patients from USA. Age 74; male 59%; ethnicity: white/Black/Asian/Hispanic 71%/16%/4%/9%; hypertension 92%; DM 25%; CHF 17%; Stroke 6%; CAD 41%; Hx AF 27%; ACEs 33%; ARBs 17%; diuretics 26%; beta blockers 62%; calcium blockers 33%; digoxin 9%; anticoagulant 23%; AADs 3% Inclusion: All patients aged >50 attending 2 outpatient cardiology clinics Exclusion: Patients with pacemakers or defibrillators	Microlife BP A 200 Omron M6 comfort	12 lead ECG, with interpretation by cardiologist/electrophysiologist
William, 2018²⁸³	52 participants from USA with 225 sets of measurements Age 68.1; 67.3% male; PAF 21.2%; persistent AF 78.8%; palpitations 42.3%; SOB 65.4%; lightheadedness 17.3%; chest pain 5.8%; fatigue 51.9% Inclusion: Patients with a diagnosis of AF admitted for AAD therapy; aged 35-85; history of PAF or persistent AF; baseline corrected QT interval <470 or 500 if QRS duration >120ms Exclusion: Patients with pacemakers; patients with defibrillators	Kardia Mobile Cardiac Monitor	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Williams, 2015²⁸⁴	99 patients from UK.29 with AF on ECG; other details not reported Inclusion: Patients attending regular AF clinic at the North west heart centre in University hospital in Manchester; Other patients attending for 12 lead ECG for reasons other than AF Exclusion: None reported	Alive Cor	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Winkler, 2011²⁸⁶	60 patients from Germany (details not provided) Inclusion: patients admitted to the cardiology department Exclusion: Not reported	Handheld ECG device with dry electrodes that records 3 lead ECG (Einthiven I, II and III leads).	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Yan, 2018²⁸⁸	233 people from Hong Kong. Mean age 70.3; 71.4% men; AF present in 34.6% at time of study; BMI 24.6; CHADSVASC 3.6; history of AF 53.9%; DM 35%; vascular disease 50.7%; TIA/stroke 18.9%; CHF 31.8%; pacemaker 3.2%; hypertension 5.9%; no antithrombotic treatment 51.2%; DOACS 13.4%; VKAs 15.7% Inclusion: Patients admitted to the cardiology ward of the hospital for clinical reasons Exclusion: None reported	Iphone units installed with Cardio Rhythm application for facial and fingertip photoplethysmographic (PPG) measures	12 lead ECG, with interpretation by cardiologist/electrophysiologist
Zwart, 2020²⁹⁵	439 patients recruited from an Outpatient Geriatric Clinic in the Netherlands. Age 78.4, female 54.4%, hypertension 63.3%, DM 22.3%, CHADSVASC 3.8, HASBLED 1.5, any stroke 15.5%, HF 11.2%, IHD 21.6% Inclusion: All consecutive patients aged ≥ 65 years at the outpatient geriatric clinic, memory clinic, or Fall and Syncope day clinic (FSC) Exclusion: Patients with pacemakers or implantable cardioverter defibrillators (ICD) or patients unable or unwilling to provide informed consent were excluded	MyDiagnostik	12 lead ECG

Table 3Summary of studies included in the evidence review for detection of atrial fibrillation

Study	Population	Index test(s)	Reference standard
Arevalo-Manso, 2016⁷	76 patients from Spain referred to a stroke centre which provides expertise to a population of about one million people, and has a dedicated SU with continuous bedside ECG monitoring for six patients. There were two samples in this study. “Study” group (n=17) were age 72.6; 47.1% men; 70.6% hypertension; 35.3% DM; 64.7% dyslipidaemia; 23.5% smokers; 35.3% CAD; 11.8% PAD; 0% TIA; 100% brain infarction; antiplatelets 52.9%; OACs 5.9%. These were assigned to one bed in the SU that was equipped with the AF-RS monitor “Control” group (n=59) were 71.9 yrs; 62.7% men; 69.55 hypertension; 25.4% DM; 61% dyslipidaemia; 20.3% smokers; 15.3% CAD; 5.1% PAD; 11.9% TIA; 88.1% brain infarction; antiplatelets 39%; OACs 3.4%. These were assigned to 5 beds in the SU that were equipped with a standard monitor Patients assigned non-randomly to these groups on basis of availability of the bed and the criteria of the neurologists on call, who were unaware of the study. Inclusion: Age>18 years and having been admitted to the SU for an acute TIA or ischaemic stroke. Exclusion: History of AF	a monitor equipped with AF-RS (DASH 5000, General Electric Healthcare, Milwaukee, Wisconsin, USA) Standard ECG monitoring devices without AF-RS.	12-lead ECG and 24 hr Holter ECG
Brown, 2019²⁴	265 patients on a stroke unit in USA. Inclusion: Ischaemic stroke or TIA in 6 bed stroke unit; 18 or over; discharged with diagnosis of acute ischaemic stroke or TIA Exclusion: Pacemaker	>24 hrs telemetry with ‘electrocardiomatrix’	>24 hours telemetry
Karunadas, 2020¹²⁶	141 patients recruited from cardiology department in India. Age of the patents ranged from 9 years to 77 years with maximum number of patients in the age group of 40-60 (mean age 44.41 years, SD 19.409). Majority were females (n ¼ 74, 52.5%). Inclusion: Patients who needed AECG monitoring as part of their clinical workup and who consented for simultaneous evaluation with the two AECG systems were included. Exclusion: Critically ill patients, those with implanted devices like permanent pacemaker or implantable cardioverter defibrillator were excluded.	Android App basedWebCardio using WiPatch for 24 hours	24 hour Holter
Kollias, 2018¹³²	100 patients attending a hypertension clinic in Greece. Age 70.6; BMI 29.1; 52.9% male; 11% stroke; 85% hypertension; 20% DM; 7% CAD; 82% antihypertensive treatment; CHADSVASC score 3.06 Inclusion: Patients attending a hypertension clinic for BP assessment, treated or untreated for hypertension; aged >=65; aged 50-64 with symptoms suggesting arrhythmias or with stroke/AF history; clinical indication for ambulatory blood pressure monitoring Exclusion: Pacemaker implantation	Microlife WatchBP O3 Afib oscillometric device with measurements programmed at 20-minute intervals for 24 hours.	24 hour Holter recording using the SpiderView (ELA Medical, Sorin Group) multichannel system recorder which was performed simultaneously with 24-hour ABPM.
Lai, 2020¹⁴⁴	40 patients recruited from a Department of Cardiovascular Ultrasound and Cardiology in China. Age 68, female 5%, AF patients (35 persistent, 2 paroxysmal and 18 SR) having prior ablation Inclusion: All consecutive patients with a history of paroxysmal or persistent AF Exclusion: Patients with pacemakers or defibrillators	Single lead (MP1) patch-based ambulatory ECG monitor worn for 24 hours; This used an automated AF detection algorithm on the basis of a convoluted neural network.	12 lead Holter ECG for 24 hours
Lyckhage, 2020¹⁶⁰	366 patients recruited from primary care in Denmark. Age 70; 34.4% female; 3.9 years since stroke; >1 clinical stroke or TIA 28.4%; CHADSVASC 4; IHD 7.4%; HF 0.3%; hypertension 69.7%; DM 13.7%; KD 3.6% Inclusion: AF-naive, had ischaemic stroke over 1 year before enrolment and were older than 49 at stroke onset. Participants with an acute infection or surgery were included at least 1 month after remission. Participants taking OAC for other indications than AF were included. Exclusion: Participants with a systemic infection or taking antiarrhythmic drugs (class I and III, digoxin, flecainide, and non-dihydropyridine calcium-channel blockers), who had cECG within 1 year before inclusion, and who had an implanted loop recorder, cardioverter defibrillator or pacemaker.	12 lead ECG, Pulse palpation – radial pulse for at least 20s.	7 day Holter
Mulder, 2012¹⁷¹	96 patients from Holland who had undergone PVI 12 months previously for paroxysmal AF; 25% female; 39% hypertension; 7% LVEF <55%; 13% mitral regurgitation grade 2; age 59; duration of AF 7 years Inclusion: Patients who had undergone PVI 12 months previously for paroxysmal AF Exclusion: None reported	Holter for 1,2,3,4,5,6 days	7 day Holter
Muller, 2009¹⁷²	48 people from Germany. Mean age 62; 29/48 male; 24 with AF; consecutive patients at an internal medicine department. Inclusion: Presence of an indication for 24 hr Holter ECG Exclusion; Antibradycardic pacemakers; implantable cardioverters and defibrillators	Vitaphone 3100 BT external loop recorder	24 hours 3 channel ECG (Holter).
Poulsen, 2017¹⁹⁶	100 patients from Denmark. Age 78; male 43/95; TIA 18/95; median CHADSVASC 5; median NIHSS 1; median time from stroke 4 days; median number of thumb ECG recordings 59; median duration of Holter monitoring 4.8 days >65 years; no history of AF who suffered an acute stroke or TIA of unknown origin in past 3 months verified by CT or MRI or clinically diagnosed; ability to handle thumb ECG None reported	30s thumb ECG (Zenicor Medical Systems AB) twice daily for 30 days	5 days Holter (Lifecard CF device).
Rizos, 2010²¹⁴	136 patients from Germany admitted to a tertiary care stroke unit; age 72; male 58.8%; manifest stroke 88.2%; TIA 11.8%; duration of bedside ECG monitoring 97hrs; CHF 36%; MI 22.8%; HT 79.4%; DM 30.1% Inclusion: Patients > 60 years presenting with an acute ischemic stroke or TIA in the ER and who were subsequently admitted to the stroke unit of our hospital and underwent continuous ECG monitoring for a minimum period of 48 h were enrolled Exclusion: Patients with AF on the initial 12-channel ECG (ELI 350; Mortara Instruments, Milwaukee, Wisc., USA) in the ER or a history of paroxysmal or persistent AF were excluded	6 channel Holter (H12+, Mortara Instruments) performed for 24 hours. 12-bit resolution digital ECG recoding for 1-2 hours.	Continuous ECG bedside monitoring for duration of stay in stroke unit
Ross, 2018²¹⁸	798 patients (409 with stroke known to be due to AF and 389 with cryptogenic stroke) from Germany. Patients with stroke due to AF: 59% female; 81 years; 5% TIA; 95% CVA; NIHSS on admission 7 Patients with cryptogenic stroke: 41% female; 68 years; 12% TIA; 88% CVA; NIHSS on admission 7 Inclusion: All patients on stroke unit – those with stroke due to known or newly diagnosed AF and those with cryptogenic stroke Exclusion: None reported	SRAclinic, Apoplex medical Technologies. Stroke Risk Analysis (SRA) – software analysis of every hourly ECG snippet of continuous (non 12 lead) ECG monitoring	Patients with stroke due to AF: repetitive 12 lead ECG Cryptogenic stroke: 24 Hour Holter
Roten, 2012²¹⁹	88 patients from Switzerland (12 patients undergoing ablation included twice, before and after ablation) – therefore 100 datasets Age 62.4; male 73%; hypertension 58%; DM 8%; IHD 18%; LVEF 60; LV diam 49mm; pre-ablation 15%; post ablation 52%; no ablation 46% Inclusion: Patients attending clinic for assessment of AF burden prior to ablation, and attending for screening post ablation; patients with known or suspected paroxysmal AF; Exclusion: Patients with persistent AF; patients unable to handle the devices independently.	7 day triggered ECG (R.Test Evolution 3).	7 day continuous Holter (Lifecard CF).
Sejr, 2019²³³	1412 patients from Denmark. 56% male; age 72.8; TIA 39.8%; Ischaemic stroke 60.2%; hypertension 58.4%; LVEF <40% 1.4%; DM 14.3%; current smoker 24.6%; OACs 0.78%; Inclusion: Acute ischaemic stroke or transient ischaemic attack (TIA) with first symptoms within 1 week, age ≥60 years, no AF on 12-lead admission ECG, no prior AF according to International Classification of Diseases codes (ICD-10) from outpatient clinic visits, hospitalisations or review of medical records, no active cancer, no implanted pacemaker, no expected low compliance or precedent participation in this study and written informed consent. Exclusion: See above	R.Test Evolution 4 (NorDiaTech, Paris, France) was device used as External loop recording (ELR).	Continuous ECG monitoring for 48 hours
Velthuis, 2013²⁶⁸	153 patients from Holland. Age 67; HT 59.5%; DM 19%; COPD 5.9%; TIA 10.5%; iCVA 7.8%; CAD 6.5%; HF 1.3%; Valve disease 6.5%; Bradytachy syndrome 0.7%; other arrhythmia 0.7% Inclusion: Consecutive patients aged >18 years admitted with a provisional diagnosis of acute ischaemic stroke Exclusion: Patients with known history of AF	24 hour external loop recorder (single channel device 3100 BT, Vitaphone, Mannheim) using automated settings	24 hour external loop recorder, interpreted by 2 blinded qualified analysts
Wasserlauf, 2019²⁷⁵	26 patients from USA with an implanted cardiac monitor. Mean age 72.1; female 34.6%; stroke 15.4%; TIA 7.7%; CHF 0%; DM 7.7%; Hypertension 69.2%; CAD 15.4%; prior MI 7.7%; CHADSVASC 2 or more 92.2%; AADs 34.6%; OACs 84.6% Inclusion: Patients with previously implanted ICMs and a history of paroxysmal AF were eligible for enrolment. Exclusion: not reported	Kardia-Band (KB; AliveCor, Mountain View, CA) - smartwatch accessory that allows a patient to record a 30-second lead I rhythm strip. Watch worn during waking hours (mean 11.3 hrs/day, over a mean of 110 days)	Insertable Cardiac Monitor
Wiesel, 2013²⁷⁸	160 patients from USA. Age 67; male 37%; white 71%; black 5%, Hispanic 5%; Asian 4%; hypertension 85%; DM 12%; CHF 6%; stroke 3%; AF 12%; CHADS2 1.4; ACEI 27%; ARB 16%; Ca channel blocker 15%; beta blocker 27%; diuretic 28%; warfarin 10% Inclusion: Patients attending general internists offices; more than or equal to 1 of the following criteria: Age >=65; hypertension, DM, CHF, stroke; patients allowed to have AF Exclusion: Pacemakers; defibrillators	AF-BP monitor for 30 days	Electrocardiographic event monitor (Hearttrack 2) [regarded as a Holter equivalent] for 30 days

Table 4Clinical evidence summary: diagnostic test accuracy for mobile ECG devices (12 lead ECG interpreted by expert cardiologist/electrophysiologist as gold standard)

Where 95% CIs are provided in round brackets (or no 95% CIs are given), raw data were not available and Forest Plots or pooled analyses were not possible.

Index Test	Number of studies	n	Interpreter of index test	Gold standard simultaneous with index test?	Sensitivity (95% CI)	Specificity (95% CI)	Risk of bias	Indirectness	Inconsistency	Imprecision	GRADE
Alive Cor handheld lead I ECG 1 measure of 30s (for Lown, 2018 it lasted a single bp cycle which is assumed to be similar)	8 Cunha, 2019⁴⁹ Desteghe, 2017⁵⁸ Desteghe, 2017⁵⁸ Haberman, 2015⁹⁰ Himmelreich, 2019¹⁰¹ Himmelreich, 2019¹⁰¹ Koltowski, 2019¹³³ Lown, 2018¹⁵⁶ William, 2018²⁸³ William, 2018²⁸³ Williams, 2015²⁸⁴ Williams, 2015²⁸⁴	1544	auto auto expert expert auto expert expert auto auto expert clinician expert	no no no no yes yes no no no no yes yes	0.91[0.71-0.99] 0.55 [0.32, 0.76] 1.00 [0.83, 1.00] 0.94 [0.73, 1.00] 0.87 [0.66-0.97] 1.00 [0.85, 1.00] 0.928^f 0.88 [0.79, 0.94] 0.71 [0.60, 0.81] 0.94 [0.86, 0.98] 0.90 [0.73, 0.98] 0.93 [0.77, 0.99] Pooled: 0.91(0.82-0.96)^e	0.97[0.91-1.00] 0.98 [0.95, 0.99] 0.98 [0.95, 0.99] 0.99 [0.95, 1.00] 0.98 [0.95, 1.00] 1.00 [0.98, 1.00] 1.000^f 0.99 [0.97, 1.00] 0.67 [0.59, 0.75] 0.87 [0.80, 0.92] 0.86 [0.76, 0.94] 0.76 [0.64, 0.85] Pooled: 0.96(0.90-0.99)^e	Sensitivity
							Very serious^a	serious^b	No serious inconsistency	Serious^c	VERY LOW
							Specificity
							Very serious^a	serious^b	No serious inconsistency	no serious imprecision	VERY LOW
	5 Desteghe, 2017⁵⁸ Lown, 2018¹⁵⁶ William, 2018²⁸³ Himmelreich, 2019¹⁰¹ Cunha, 2019⁴⁹		AUTO SUBGROUP		0.55 [0.32, 0.76] 0.88 [0.79, 0.94] 0.71 [0.60, 0.81] 0.87 [0.66-0.97] 0.91[0.71-0.99] Pooled: 0.81(0.61-0.92)^e	0.98 [0.95, 0.99] 0.99 [0.97, 1.00] 0.67 [0.59, 0.75] 0.98 [0.95, 1.00] 0.97 [0.91-1.00] Pooled: 0.96(0.83-0.99)^e	Sensitivity
							Very serious^a	serious^b	No serious inconsistency	Serious^c	VERY LOW
							Specificity
							Very serious^a	serious^b	No serious inconsistency	Serious^c	VERY LOW
	6 Desteghe, 2017⁵⁸ Haberman, 2015⁹⁰ Williams, 2015²⁸⁴ Koltowski, 2019¹³³ William, 2018²⁸³ Himmelreich, 2019¹⁰¹		EXPERT SUBGROUP		1.00 [0.83, 1.00] 0.94 [0.73, 1.00] 0.93 [0.77, 0.99] 0.928^f 0.94 [0.86, 0.98] 1.00 [0.85, 1.00] Pooled: 0.95(0.88-0.99)^e	0.98 [0.95, 0.99] 0.99 [0.95, 1.00] 0.76 [0.64, 0.85] 1.000^f 0.87 [0.80, 0.92] 1.00 [0.98, 1.00] Pooled: 0.96(0.81-0.99)^e	Sensitivity
							Very serious^a	serious^b	No serious inconsistency	Serious^c	VERY LOW
							Specificity
							Very serious^a	serious^b	No serious inconsistency	Serious^c	VERY LOW
	1 Williams, 2015²⁸⁴		CLINICIAN SUBGROUP		0.90 [0.73, 0.98]	0.86 [0.76, 0.94]	Sensitivity
							serious^a	serious^b	Not applicable	Serious^c	VERY LOW
							Specificity
							serious^a	serious^b	Not applicable	Serious^c	VERY LOW
Kardia band (Alive Cor) watch device (equivalent to Lead I) 1 measure for 30s	2 Bumgarner, 2018²⁶ Bumgarner, 2018²⁶ Rajakariar, 2020²⁰¹	369	auto expert auto	no no no	0.69 [0.59, 0.78] 0.88 [0.79, 0.94] 0.94[0.82-0.99] Pooled: 0.86[0.50-0.97]	0.91 [0.82, 0.96] 0.86 [0.76, 0.93] 0.82[0.75-0.88] Pooled: 0.87[0.65-0.96]	Sensitivity
							serious^a	serious^b	Serious^d	Very serious^c	VERY LOW
							Specificity
							serious^a	serious^b	No serious inconsistency	Serious^c	VERY LOW
Beurer ME90 device – lead I ECG 1 measure of 30s	1 Brito, 2018²³ Brito, 2018²³	126	auto expert	no no	0.89 [0.65, 0.99] 0.84 [0.60, 0.97] Median^g: 0.84[0.60, 0.97]	0.62 [0.52, 0.71] 1.00 [0.97, 1.00] 1.00[0.97, 1.00]	Sensitivity
							Very serious^a	serious^b	No serious inconsistency	Serious^c	VERY LOW
							Specificity
							Very serious^a	serious^b	Serious^d	No serious imprecision	VERY LOW
Beurer ME90 device – lead I and mv4 leads 1 measure of 30s	1 Brito, 2018²³	126	auto	no	0.88 [0.64, 0.99]	0.84 [0.76, 0.91]	Sensitivity
							Very serious^a	serious^b	Not applicable	Serious^c	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	Serious^c	VERY LOW
Beurer ME90 device – mv4 lead 1 measure of 30s	1 Brito, 2018²³ Brito, 2018²³	126	auto expert	no no	0.94 [0.71, 1.00] 0.84 [0.60, 0.97] Median^g: 0.84[0.60, 0.97]	0.76 [0.67, 0.84] 1.00 [0.97, 1.00] 1.00 [0.97, 1.00]	Sensitivity
							Very serious^a	serious^b	No serious inconsistency	Serious^c	VERY LOW
							Specificity
							Very serious^a	serious^b	Serious^d	no serious imprecision	VERY LOW
Beurer ME 80 device – a pocket sized (reconstructing 9 lead) ECG device. 1 measure of unknown duration	1 Nigolian, 2018¹⁷⁷ Nigolian, 2018¹⁷⁷	52	clinician expert	no no	0.75 [0.48, 0.93] 1.00 [0.79, 1.00] Median^g: 0.75 [0.48, 0.93]	0.89 [0.74, 0.97] 0.94 [0.81, 0.99] 0.89 [0.74, 0.97]	Sensitivity
							serious^a	serious^b	No serious inconsistency	Very serious^c	VERY LOW
							Specificity
							serious^a	serious^b	No serious inconsistency	Serious^c	VERY LOW
ECG Check, an FDA-approved mobile heart monitor manufactured by Cardiac Designs. By putting two fingers on the ECG Check, it registers a lead I ECG 1 measure of 30s	1 Haverkamp, 2019⁹⁶	94	auto	no	1.00 [0.72, 1.00]	0.94 [0.86, 0.98]	Sensitivity
							Very serious^a	serious^b	Not applicable	Serious^c	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	Serious^c	VERY LOW
Merlin ECG event recorder (single lead) 1 measure of 30s	1 Kearley, 2014¹²⁸	1000	expert	no	0.939^f	0.901^f	Sensitivity
							serious^a	serious^b	Not applicable	Not assessed	VERY LOW
							Specificity
							serious^a	serious^b	Not applicable	Not assessed	VERY LOW
My Diagnostik handheld lead I ECG 1 measure of 60s	2 Desteghe, 2017⁵⁸ Desteghe, 2017⁵⁸ Tieleman, 2014²⁵⁸	1125	auto expert auto	no no no	0.82 [0.60, 0.95] 0.85 [0.62, 0.97] 1.00 [0.97, 1.00] Pooled:0.94(0.52-0.99)^e	0.94 [0.91, 0.97] 0.95 [0.92, 0.98] 0.98 [0.97, 0.99] Pooled:0.97(0.85-0.99)^e	Sensitivity
							serious^a	serious^b	No serious inconsistency	Very serious^c	VERY LOW
							Specificity
							serious^a	serious^b	No serious inconsistency	Serious^c	VERY LOW
My Diagnostik handheld lead I ECG 3 measures of 60s (majority rule)	1 Vaes, 2014²⁶⁵	181	auto	no	0.94 [0.87, 0.98]	0.93 [0.85, 0.97]	Sensitivity
							Very serious^a	serious^b	Not applicable	Serious^c	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	Serious^c	VERY LOW
My Diagnostik handheld lead I ECG 3 measures of unknown time on 3 different occasions (unclear intervals)	1 Zwart, 2020²⁹⁵	439	expert	no	0.90^f	0.99^f	Sensitivity
							Very serious^a	serious^b	Not applicable	Not assessed	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	Not assessed	VERY LOW
Omron Heartscan HCG 801 E single lead device 1 measure of 30s (though duration not stated in Kearley, 2014¹²⁸)	3 Kearley, 2014¹²⁸ Kearley, 2014¹²⁸ Renier, 2012²⁰⁸ Renier, 2012²⁰⁸ Kaleschke, 2009¹¹⁷	1684	auto expert auto clinician expert	no no no no no	0.99 [0.93, 1.00] 0.944^f 0.92 [0.64, 1.00] 0.69 [0.39, 0.91] 0.99(0.96-1.00)^f Pooled:0.93(0.50-0.99)^e	0.76 [0.73, 0.79] 0.946^f 1.00 [0.98, 1.00] 0.95 [0.90, 0.97] 0.96(0.94-0.98)^f Pooled:0.95(0.5 2-0.99)^e	Sensitivity
							serious^a	serious^b	No serious inconsistency	Very serious^c	VERY LOW
							Specificity
							serious^a	serious^b	Serious^d	Very serious^c	VERY LOW
ECG Bone – single lead Unclear reps and duration	1 Proesmans, 2019¹⁹⁷	223	expert	no	0.90(0.824-0.951)	0.968(0.919-0.991)	Sensitivity
							Very serious^a	serious^b	Not applicable	Serious^c	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	no serious imprecision	VERY LOW
Zenicor ECG thumb device (bipolar lead I) 1 measure of 10s	1 Doliwa, 2009⁶³	100	expert	no	0.96 [0.86, 1.00]	0.92 [0.81, 0.98]	Sensitivity
							serious^a	serious^b	Not applicable	Serious^c	VERY LOW
							Specificity
							serious^a	serious^b	Not applicable	Serious^c	VERY LOW
Polar H7 heart rate monitor – heart rate sensor derives ECG data via chest electrodes. Can detect RR intervals accurately 1 measure over duration of a bp measurement cycle	1 Lown, 2018¹⁵⁶	418	auto	no	0.96 [0.90, 0.99]	0.98 [0.96, 0.99]	Sensitivity
							Very serious^a	serious^b	Not applicable	no serious imprecision	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	no serious imprecision	VERY LOW
RITMIA HR monitor 1 measure over 10 minutes	1 Reverberi, 2019²⁰⁹	100	auto	no	0.97 [0.91, 0.99]	0.96 [0.89, 0.99]	Sensitivity
							serious^a	serious^b	Not applicable	No serious imprecision	LOW
							Specificity
							serious^a	serious^b	Not applicable	Serious^c	VERY LOW
Firstbeat Bodyguard 2 – delivers single lead data 1 measure over duration of a bp measurement cycle	1 Lown, 2018¹⁵⁶	418	auto	no	0.96 [0.90, 0.99]	0.99 [0.97, 1.00]	Sensitivity
							Very serious^a	serious^b	Not applicable	no serious imprecision	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	no serious imprecision	VERY LOW
Cardiobip, a portable 5 lead handheld ECG system for remote monitoring of patients. 1-3 measures of unknown duration	1 Vukajlovic, 2010²⁷¹	36	expert	no	1.00 [0.85, 1.00]	1.00 [0.77, 1.00]	Sensitivity
							serious^a	serious^b	Not applicable	Serious^c	VERY LOW
							Specificity
							serious^a	serious^b	Not applicable	Serious^c	VERY LOW
Mobile ECG device ER-2000s. Mode 1 uses three ECG electrodes that are attached to the anterior chest wall. patients were instructed to push the record button when they believed they were experiencing a cardiac symptom. 1 measure of unknown duration	1 Park, 2015¹⁸⁶	17	expert	yes	1.000^f	1.000^f	Sensitivity
							Very serious^a	No serious indirectness	Not applicable	Not assessed	LOW
							Specificity
							Very serious^a	No serious indirectness	Not applicable	Not assessed	LOW
Mobile ECG device ER-2000s. Mode 2 uses the side chest channel and finger channel. Patients were instructed to push the record button when they believed they were experiencing a cardiac symptom. 1 measure of unknown duration	1 Park, 2015¹⁸⁶	17	expert	yes	1.000^f	1.000^f	Sensitivity
							Very serious^a	No serious indirectness	Not applicable	Not assessed	LOW
							Specificity
							Very serious^a	No serious indirectness	Not applicable	Not assessed	LOW
Huawei band 2 smartband 1 measure of 3mins	1 Fan, 2019⁷⁷	624	auto	yes	0.92 [0.88, 0.95]	1.00 [0.98, 1.00]	Sensitivity
							serious^a	serious^b	Not applicable	Serious^c	VERY LOW
							Specificity
							serious^a	serious^b	Not applicable	No serious imprecision	LOW
Amazfit Healthband - ECG 1 measure of 60 seconds	1 Chen, 2020³⁶	401	auto	no	0.87 [0.81, 0.92]	1.00 [0.99, 1.00]	Sensitivity
							serious^a	No serious indirectness	Not applicable	Serious^c	LOW
							Specificity
							serious^a	No serious indirectness	Not applicable	No serious imprecision	MODERATE
Atrial Fibrillation Screening Device 1 measure of 60 seconds	1 Diamantino, 2020⁵⁹	334	auto	no	0.90 [0.77, 0.97]	0.84 [0.79, 0.88]	Sensitivity
							serious^a	No serious indirectness	Not applicable	Serious^c	LOW
							Specificity
							serious^a	No serious indirectness	Not applicable	No serious imprecision	MODERATE
Rithmi heart rhythm monitor: wrist monitor using ECG lead 1 measure of 3 minutes	1 Osca Asensi, 2020¹⁸⁴	167	auto	no	0.94^f	0.96^f	Sensitivity
							Very serious^a	serious^b	Not applicable	Not assessed	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	Not assessed	VERY LOW
Snap ECG – portable single lead 1 measure of 60 seconds	1 Guan, 2020⁸⁶	1479	expert	no	0.65 [0.41, 0.85]	0.99 [0.99, 1.00]	Sensitivity
							serious^a	Serious^b	Not applicable	Serious^c	VERY LOW
							Specificity
							serious^a	Serious^b	Not applicable	No serious imprecision	LOW

(a): Risk of bias was assessed using the QUADAS-2 checklist. The evidence was downgraded by 1 increment if the majority of studies were rated at high risk of bias, and downgraded by 2 increments if the majority of studies were rated at very high risk of bias.

(b): Indirectness was assessed using the QUADAS-2 checklist items referring to applicability. The evidence was downgraded by 1 increment if the majority of studies were seriously indirect, and downgraded by 2 increments if the majority of studies are very seriously indirect

(c): Imprecision was assessed based on inspection of the confidence region in the diagnostic meta-analysis or, where diagnostic meta-analysis has not been conducted, assessed according to the range of confidence intervals in the individual studies. The evidence was downgraded by 1 increment when the confidence interval around the point estimate crossed one of the clinical thresholds (0.90 or 0.60), and downgraded by 2 increments when the confidence interval around the point estimate crossed both of the clinical thresholds. The upper clinical threshold marked the point above which recommendations would be possible, and the lower clinical threshold marked the point below which the tool would be regarded as of little clinical use.

(d): Inconsistency was assessed by visual inspection of the sensitivity/specificity plots, or data (if 2 studies). The evidence was downgraded by 1 increment if there was no overlap of 95% confidence intervals. For single studies no evaluation was made and ‘not applicable’ was recorded.

(e): Pooled sensitivity/specificity from diagnostic meta-analysis

(f): indicates that because the raw data were not provided by the paper, and were also not able to be calculated by the reviewer, these were not able to be included in any diagnostic meta-analyses

(g): For non-pooled analyses, the median sensitivity and the paired specificity were reported. If there were an even number of studies the lower middle value was given

Table 5Clinical evidence summary: diagnostic test accuracy for blood pressure monitors (12 lead ECG interpreted by expert cardiologist/electrophysiologist as gold standard)

Where 95% CIs are provided in round brackets (or no 95% CIs are given), raw data were not available and Forest Plots or pooled analyses were not possible.

Index Test	Number of studies	n	Interpreter of index test	Gold standard simultaneous with index test?	Sensitivity (95% CI)	Specificity (95% CI)	Risk of bias	Indirectness	nconsistency	Imprecision	GRADE
Microlife BP3MQ1-2D oscillometric device 3 readings, with ‘majority rule’ of 10 beat intervals	2 Gandolfo, 2015⁷⁹ Wiesel, 2009²⁸⁰	612	auto auto	no no	0.89 [0.75, 0.97] 0.97 [0.91, 0.99] Median^g: 0.89 [0.75, 0.97]	0.99 [0.96, 1.00] 0.89 [0.85, 0.92] 0.99 [0.96, 1.00]	Sensitivity
							serious^a	serious^b	No serious inconsistency	Serious^d	VERY LOW
							Specificity
							serious^a	serious^b	serious^c	No serious imprecision	VERY LOW
Microlife BP3MQ1-2D oscillometric device Single readings of 10 beat intervals	1 Wiesel, 2009²⁸⁰	1215	auto	no	0.95 [0.92, 0.97]	0.86 [0.84, 0.89]	Sensitivity
							serious^a	serious^b	Not applicable	No serious imprecision	LOW
							Specificity
							serious^a	serious^b	Not applicable	No serious imprecision	LOW
Microlife BPA 200 oscillometric device 3 readings, with majority rule of 10 beat intervals	2 Marazzi, 2012¹⁶² Wiesel, 2014²⁷⁹	686	auto auto	yes no	0.92 [0.85, 0.97] 1.00 [0.88, 1.00] Median^g: 0.92 [0.85, 0.97]	0.97 [0.95, 0.98] 0.92 [0.87, 0.96] 0.97 [0.95, 0.98]	Sensitivity
							Serious^a	No serious indirectness	No serious inconsistency	serious^d	LOW
							Specificity
							Serious^a	No serious indirectness	No serious inconsistency	No serious imprecision	MOD
Microlife BPA100 Plus BP oscillometric device 3 readings (majority rule) of 10 beat intervals	1 Stergio, 2009²⁴³	72	auto	yes	1.00 [0.87, 1.00]	0.89 [0.76, 0.96]	Sensitivity
							Serious^a	No serious indirectness	Not applicable	serious^d	LOW
							Specificity
							Serious^a	No serious indirectness	Not applicable	serious^d	LOW
Microlife BPA100 Plus BP oscillometric device 3 readings (minority rule) of 10 beat intervals	1 Stergio, 2009²⁴³	72	auto	yes	1.00 [0.87, 1.00]	0.69 [0.53, 0.82]	Sensitivity
							Serious^a	No serious indirectness	Not applicable	serious^d	LOW
							Specificity
							Serious^a	No serious indirectness	Not applicable	serious^d	LOW
Microlife BPA100 Plus BP oscillometric device 1^st reading only of 10 beat intervals	1 Stergio, 2009²⁴³	72	auto	yes	0.93 [0.76, 0.99]	0.89 [0.76, 0.96]	Sensitivity
							Serious^a	No serious indirectness	Not applicable	serious^d	LOW
							Specificity
							Serious^a	No serious indirectness	Not applicable	serious^d	LOW
Microlife BPA100 Plus BP oscillometric device 1^st 2 readings of 10 beat intervals	1 Stergio, 2009²⁴³	72	auto	yes	1.00 [0.87, 1.00]	0.76 [0.60, 0.87]	Sensitivity
							Serious^a	No serious indirectness	Not applicable	serious^d	LOW
							Specificity
							Serious^a	No serious indirectness	Not applicable	No serious imprecision	MOD
Microlife Watch BP oscillometric device Readings over up to 3 bp measurement cycles	2 Kearley, 2014¹²⁸ Lown, 2018¹⁵⁶	1417	auto auto	no no	0.95 [0.88, 0.99] 0.96 [0.90, 0.99] Median^g: 0.95 [0.88, 0.99]	0.90 [0.88, 0.92] 0.93 [0.90, 0.96] 0.90 [0.88, 0.92]	Sensitivity
							Very serious^a	serious^b	No serious inconsistency	serious^d	VERY LOW
							Specificity
							Very serious^a	serious^b	No serious inconsistency	serious^d	VERY LOW
Heart Spectrum Blood Pressure Monitor algorithm 1 (see evidence tables) 1 reading(unknown duration)	1 Kao, 2018¹²³	62	expert	yes	0.97 [0.82, 1.00]	0.97 [0.84, 1.00]	Sensitivity
							Very serious^a	serious^b	Not applicable	serious^d	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	serious^d	VERY LOW
Heart Spectrum Blood Pressure Monitor algorithm 2 (see evidence tables) 1 reading(unknown duration)	1 Kao, 2018¹²³	62	expert	yes	0.90 [0.73, 0.98]	1.00 [0.89, 1.00]	Sensitivity
							Very serious^a	serious^b	Not applicable	serious^d	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	serious^d	VERY LOW
Heart Spectrum Blood Pressure Monitor algorithm 3 (see evidence tables) 1 reading(unknown duration)	1 Kao, 2018¹²³	62	expert	yes	1.00 [0.88, 1.00]	0.94 [0.80, 0.99]	Sensitivity
							Very serious^a	serious^b	Not applicable	serious^d	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	serious^d	VERY LOW
Omron 712C automatic sphygmomanometer 2 readings of 10-40 secs each	1 Wiesel, 2004²⁸¹	450	auto	no	1.00 [0.93, 1.00]	0.91 [0.88, 0.94]	Sensitivity
							Very serious^a	No serious indirectness	Not applicable	No serious imprecision	LOW
							Specificity
							Very serious^a	No serious indirectness	Not applicable	serious^d	VERY LOW
Omron M6 Comfort 1 reading(unknown duration)	2 Marazzi, 2012¹⁶² Wiesel, 2014²⁷⁹	686	yes no	auto auto	1.00 [0.96, 1.00] 0.33 [0.17, 0.53] Median^g: 0.33 [0.17, 0.53]	0.94 [0.92, 0.96] 0.97 [0.93, 0.99] 0.97 [0.93, 0.99]	Sensitivity
							Serious^a	No serious indirectness	serious^c	no serious imprecision	LOW
							Specificity
							Serious^a	No serious indirectness	No serious imprecision	no serious imprecision	MOD

(a): Risk of bias was assessed using the QUADAS-2 checklist. The evidence was downgraded by 1 increment if the majority of studies were rated at high risk of bias, and downgraded by 2 increments if the majority of studies were rated at very high risk of bias.

(b): Indirectness was assessed using the QUADAS-2 checklist items referring to applicability. The evidence was downgraded by 1 increment if the majority of studies were seriously indirect, and downgraded by 2 increments if the majority of studies are very seriously indirect

(c): Inconsistency was assessed by visual inspection of the sensitivity/specificity plots, or data (if 2 studies). The evidence was downgraded by 1 increment if there was no overlap of 95% confidence intervals. For single studies no evaluation was made and ‘not applicable’ was recorded.

(d): Imprecision was assessed based on inspection of the confidence region in the diagnostic meta-analysis or, where diagnostic meta-analysis has not been conducted, assessed according to the range of confidence intervals in the individual studies. The evidence was downgraded by 1 increment when the confidence interval around the point estimate crossed one of the clinical thresholds (0.90 or 0.60), and downgraded by 2 increments when the confidence interval around the point estimate crossed both of the clinical thresholds. The upper clinical threshold marked the point above which recommendations would be possible, and the lower clinical threshold marked the point below which the tool would be regarded as of little clinical use.

(e): Pooled sensitivity/specificity from diagnostic meta-analysis

(f): indicates that because the raw data were not provided by the paper, and were also not able to be calculated by the reviewer, these were not able to be included in any diagnostic meta-analyses

(g): For non-pooled analyses, the median sensitivity and the paired specificity were reported. If there were an even number of studies the lower middle value was given.

Table 6Clinical evidence summary: diagnostic test accuracy for pulse palpation (12 lead ECG interpreted by expert cardiologist/electrophysiologist as gold standard)

Where 95% CIs are provided in round brackets (or no 95% CIs are given), raw data were not available and Forest Plots or pooled analyses were not possible.

Index Test	Number of studies	n	Interpreter of index test	Gold standard simultaneous with index	Sensitivity (95% CI)	Specificity (95% CI)	Risk of bias	Indirectness	Inconsistency	Imprecsion	GRADE
Pulse palpation 1 reading(unknown duration)	2 Hobbs, 2005¹⁰⁴ Somerville, 2000²⁴⁰ Somerville, 2000²⁴⁰	2616	Clinician Clinician expert	yes no no	0.87 [0.82, 0.91] 0.92 [0.64, 1.00] 1.00 [0.87, 1.00] Pooled:0.92(0.71-0.99)^d	0.81 [0.80, 0.83] 0.84 [0.64, 0.95] 0.77 [0.64, 0.87] Pooled:0.81(0.56-0.94)^d	Sensitivity
							serious^a	serious^b	No serious inconsistency	Serious^c	VERY LOW
							Specificity
							serious^a	serious^b	No serious inconsistency	very serious^d	VERY LOW

(a): Risk of bias was assessed using the QUADAS-2 checklist. The evidence was downgraded by 1 increment if the majority of studies were rated at high risk of bias, and downgraded by 2 increments if the majority of studies were rated at very high risk of bias.

(b): Indirectness was assessed using the QUADAS-2 checklist items referring to applicability. The evidence was downgraded by 1 increment if the majority of studies were seriously indirect, and downgraded by 2 increments if the majority of studies are very seriously indirect

(c): Imprecision was assessed based on inspection of the confidence region in the diagnostic meta-analysis or, where diagnostic meta-analysis has not been conducted, assessed according to the range of confidence intervals in the individual studies. The evidence was downgraded by 1 increment when the confidence interval around the point estimate crossed one of the clinical thresholds (0.90 or 0.60), and downgraded by 2 increments when the confidence interval around the point estimate crossed both of the clinical thresholds. The upper clinical threshold marked the point above which recommendations would be possible, and the lower clinical threshold marked the point below which the tool would be regarded as of little clinical use.

(d): Pooled sensitivity/specificity from diagnostic meta-analysis

Table 7Clinical evidence summary: diagnostic test accuracy for photoplethysmography (12 lead ECG interpreted by expert cardiologist/electrophysiologist as gold standard)

Index Test	Number of studies	n	Interpreter of index test	Gold standard simultaneous with index test?	Sensitivity (95% CI)	Specificity (95% CI)	Risk of bias	Indirectness	Inconsistency	Imprecsion	GRADE
iPhone 4s app - 2 minute pulse waveforms with PULSESMART app (using RMSSD, ShE and Poincare thresholds) from fingertip pulse recordings 1 reading (2 mins)	1 McManus, 2016¹⁶⁴	189	auto	yes	0.97 [0.91, 0.99]	0.93 [0.86, 0.98]	Sensitivity
							serious^a	serious^b	Not applicable	No serious imprecision	LOW
							Specificity
							serious^a	serious^b	Not applicable	Serious^c	VERY LOW
iPhone 4S camera using both RMSSD and Shannon entropy thresholds from fingertip pulse recordings 1 reading (2 mins)	1 McManus, 2013¹⁶⁵	76	Expert, calculated algorithm	no	0.962^d	0.975^d	Sensitivity
							Very serious^a	serious^b	Not applicable	Not assessed	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	Not assessed	VERY LOW
iPhone 4S camera using just RMSSD threshold from fingertip pulse recordings 1 reading (2 mins)	1 McManus, 2013¹⁶⁵	76	Expert, calculated algorithm	no	0.982^d	0.915^d	Sensitivity
							Very serious^a	serious^b	Not applicable	Not assessed	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	Not assessed	VERY LOW
iPhone 4S camera using just Shannon entropy threshold from fingertip pulse recordings 1 reading (2 mins)	1 McManus, 2013¹⁶⁵	76	Expert, calculated algorithm	no	0.975^d	0.822^d	Sensitivity
							Very serious^a	serious^b	Not applicable	Not assessed	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	Not assessed	VERY LOW
Fingertip Cardio Rhythm Mobile Application (CRMA) iphone 3 × 20s readings (majority rule)	1 Rozen, 2018²²⁰	189	auto	no	0.93[0.86, 0.97]	0.91[0.83, 0.96]	Sensitivity
							serious^a	serious^b	Not applicable	Serious^c	VERY LOW
							Specificity
							serious^a	serious^b	Not applicable	Serious^c	VERY LOW
Fingertip Cardio Rhythm Mobile Application (CRMA) iphone 3 × 20s readings (minority rule)	1 Yan, 2018²⁸⁸	217	auto	no	0.95 [0.87, 0.99]	0.93 [0.87, 0.97]	Sensitivity
							serious^a	serious^b	Not applicable	Serious^c	VERY LOW
							Specificity
							serious^a	serious^b	Not applicable	Serious^c	VERY LOW
Facial Cardio Rhythm Mobile Application (CRMA) iphone 3 × 20s readings (minority rule)	1 Yan, 2018²⁸⁸	217	auto	no	0.95 [0.87, 0.99]	0.96 [0.91, 0.98]	Sensitivity
							serious^a	serious^b	Not applicable	Serious^c	VERY LOW
							Specificity
							serious^a	serious^b	Not applicable	No serious imprecision	LOW
FibriCheck app. A PPG signal (fingertip) was acquired with the rear camera of an iPhone 5S. 3 × 1min readings	1 Proesmans, 2019¹⁹⁷	214	auto	no	0.87 [0.79, 0.93]	0.97 [0.92, 0.99]	Sensitivity
							serious^a	serious^b	Not applicable	Serious^c	VERY LOW
							Specificity
							serious^a	serious^b	Not applicable	No serious imprecision	LOW
Huawei Honor 7A fingertip/LED device (auto) 1 × 3mins readings	1 Fan, 2019⁷⁷	108	auto	yes	0.956(0.902-0.982) No raw data in paper – CIs provided by paper)	0.99.4(0.962-1.00) No raw data in paper – CIs provided by paper)	Sensitivity
							serious^a	serious^b	Not applicable	No serious imprecision	LOW
							Specificity
							serious^a	serious^b	Not applicable	No serious imprecision	LOW
Huawei Mate 9 fingertip/LED device (auto) 3mins 1 × 3mins readings	1 Fan, 2019⁷⁷	108	auto	yes	0.944(0.889-0.974) No raw data in paper – CIs provided by paper)	1.00(0.972-1.00) No raw data in paper – CIs provided by paper)	Sensitivity
							serious^a	serious^b	Not applicable	Serious^c	VERY LOW
							Specificity
							serious^a	serious^b	Not applicable	No serious imprecision	LOW
The screening technique involves a finger-probe instrument (as used in pulse oximetry) that utilises the principle of photoplethysmography. 1 × 30s reading	1 Lewis, 2011¹⁵⁰	594	auto	no	1.00^d	Incorrect calculation in paper but insufficient data to correct it	Sensitivity
							serious^a	serious^b	Not applicable	Not assessed	LOW
							Specificity
							Not applicable	Not applicable	Not applicable	Not applicable	Not applicable
Wrist-type photoplethysmographic (PPG) device. Using inter-beat interval (IBI) features (mean, SD, median, IQR, min, max and RMSSD. 1 reading (unknown duration)	1 Fallet, 2019⁷⁶	17	auto	yes	0.995^d	0.895^d	Sensitivity
							Very serious^a	serious^b	Not applicable	Not assessed	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	Not assessed	VERY LOW
Wrist-type photoplethysmographic (PPG) device. Using ‘wave’ features (Adaptive organisation Index, variance of the slope of the phase difference, permutation entropy, fractional spectral radius and spectral purity index) 1 reading (unknown duration)	1 Fallet, 2019⁷⁶	17	auto	yes	0.992^d	0.906^d	Sensitivity
							Very serious^a	serious^b	Not applicable	Not assessed	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	Not assessed	VERY LOW
Wrist-type photoplethysmographic (PPG) device. Using BOTH IBI and wave features 1 reading (unknown duration)	1 Fallet, 2019⁷⁶	17	auto	yes	0.997^d	0.924^d	Sensitivity
							Very serious^a	serious^b	Not applicable	Not assessed	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	Not assessed	VERY LOW
Amazfit Healthband - PPG 1 measure of 60 seconds	1 Chen, 2020³⁶	401	auto	no	0.88 [0.82, 0.93]	0.99 [0.97, 1.00]	Sensitivity
							serious^a	No serious indirectness	Not applicable	Serious^c	LOW
							Specificity
							serious^a	No serious indirectness	Not applicable	No serious imprecision	MODERATE
Rithmi heart rhythm monitor: wrist monitor using PPG 1 measure of 3 minutes	1 Osca Asensi, 2020¹⁸⁴	167	auto	no	0.91^d	0.96^d	Sensitivity
							Very serious^a	serious^b	Not applicable	Not assessed	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	Not assessed	VERY LOW

(a): Risk of bias was assessed using the QUADAS-2 checklist. The evidence was downgraded by 1 increment if the majority of studies were rated at high risk of bias, and downgraded by 2 increments if the majority of studies were rated at very high risk of bias.

(b): Indirectness was assessed using the QUADAS-2 checklist items referring to applicability. The evidence was downgraded by 1 increment if the majority of studies were seriously indirect, and downgraded by 2 increments if the majority of studies are very seriously indirect

(c): Imprecision was assessed based on inspection of the confidence region in the diagnostic meta-analysis or, where diagnostic meta-analysis has not been conducted, assessed according to the range of confidence intervals in the individual studies. The evidence was downgraded by 1 increment when the confidence interval around the point estimate crossed one of the clinical thresholds (0.90 or 0.60), and downgraded by 2 increments when the confidence interval around the point estimate crossed both of the clinical thresholds. The upper clinical threshold marked the point above which recommendations would be possible, and the lower clinical threshold marked the point below which the tool would be regarded as of little clinical use.

(d): indicates that because the raw data were not provided by the paper, and were also not able to be calculated by the reviewer, these were not able to be included in any diagnostic meta-analyses

Table 8Clinical evidence summary: diagnostic test accuracy for 3-lead tele ECG (12 lead ECG interpreted by expert cardiologist/electrophysiologist as gold standard)

Where 95% CIs are provided in round brackets (or no 95% CIs are given), raw data were not available and Forest Plots or pooled analyses were not possible.

Index Test	Number of studies	n	Interpreter of index test	Gold standard simultaneous with index test?	Sensitivity (95% CI)	Specificity (95% CI)	Risk of bias	Indirectness	Inconsistency	Imprecsion	GRADE
CG-7100, Card Guard Scientific Survival Ltd 3-lead tele-ECG. 2 readings (unknown duration)	1 Antonicelli, 2012⁶	107	expert	no	1.00 [0.03, 1.00]	1.00 [0.97, 1.00]	Sensitivity
							serious^a	serious^b	Not applicable	Very serious^d	VERY LOW
							Specificity
							serious^a	serious^b	Not applicable	No serious imprecision	LOW
Handheld tele ECG device with dry electrodes that records 3 lead ECG. 1 reading of 120s	1 Winkler, 2011²⁸⁶	60	automated	no	0.929^f	0.909^f	Sensitivity
							Very serious^a	serious^b	Not applicable	Not assessed	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	Not assessed	VERY LOW
Portable 3-lead ECG monitor (PEM) 1 reading of 30s	1 Kristensen, 2016¹³⁸	89	clinician	yes	0.87 [0.60, 0.98]	0.99 [0.93, 1.00]	Sensitivity
							No serious risk of bias	serious^b	Not applicable	serious^d	LOW
							Specificity
							No serious risk of bias	serious^b	Not applicable	No serious imprecision	MOD
Wearable and wireless 3-lead ECG device (Medi-Trace 200, Kendall) connected to the user via disposable button electrodes. Signals then analysed according to algorithm 1: if the variation of consecutive R-R intervals is >150ms within 6 secs of computation 1 reading of 6 min	1 Lin, 2010¹⁵³	1200	automated	no	0.93 [0.92, 0.95]^**	0.00 [0.00, 0.37]^**	Sensitivity
							Very serious^a	serious^b	Not applicable	No serious imprecision	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	No serious imprecision	VERY LOW
Wearable and wireless 3-lead ECG device (Medi-Trace 200, Kendall) connected to the user via disposable button electrodes. Signals then analysed according to algorithm 2: if the variation of consecutive R-R intervals is >150ms AND SD of R-R intervals in each 6 second recording is >60 ms within 6 seconds of computation 1 reading of 6min	1 Lin, 2010¹⁵³	1200	automated	no	0.95 [0.94, 0.96]^**	0.00 [0.00, 0.41]^**	Sensitivity
							Very serious^a	serious^b	Not applicable	No serious imprecision	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	No serious imprecision	VERY LOW

(a): Risk of bias was assessed using the QUADAS-2 checklist. The evidence was downgraded by 1 increment if the majority of studies were rated at high risk of bias, and downgraded by 2 increments if the majority of studies were rated at very high risk of bias.

(b): Indirectness was assessed using the QUADAS-2 checklist items referring to applicability. The evidence was downgraded by 1 increment if the majority of studies were seriously indirect, and downgraded by 2 increments if the majority of studies are very seriously indirect

(c): Inconsistency was assessed by visual inspection of the sensitivity/specificity plots, or data (if 2 studies). The evidence was downgraded by 1 increment if there was no overlap of 95% confidence intervals. For single studies no evaluation was made and ‘not applicable’ was recorded.

(d): Imprecision was assessed based on inspection of the confidence region in the diagnostic meta-analysis or, where diagnostic meta-analysis has not been conducted, assessed according to the range of confidence intervals in the individual studies. The evidence was downgraded by 1 increment when the confidence interval around the point estimate crossed one of the clinical thresholds (0.90 or 0.60), and downgraded by 2 increments when the confidence interval around the point estimate crossed both of the clinical thresholds. The upper clinical threshold marked the point above which recommendations would be possible, and the lower clinical threshold marked the point below which the tool would be regarded as of little clinical use.

(e): Pooled sensitivity/specificity from diagnostic meta-analysis

(f): indicates that because the raw data were not provided by the paper, and were also not able to be calculated by the reviewer, these were not able to be included in any diagnostic meta-analyses

(g): For non-pooled analyses, the median sensitivity and the paired specificity were reported. If there were an even number of studies the lower middle value was given.

**: unit of analysis was each of the separate measures done on each person

Table 9Clinical evidence summary: diagnostic test accuracy for 6 lead ECG (12 lead ECG interpreted by expert cardiologist/electrophysiologist as gold standard)

Where 95% CIs are provided in round brackets (or no 95% CIs are given), raw data were not available and Forest Plots or pooled analyses were not possible.

Index Test	Number of studies	n	Interpreter of index test	Gold standard simultaneous with index test?	Sensitivity (95% CI)	Specificity (95% CI)	Risk of bias	Indirectness	Inconsistency	Imprecsion	GRADE
6 lead ECG with prototype recorder placed on thorax/abdomen in sitting with loosed clothing only 1 measure of 5 seconds	1 Caldwell, 2012²⁹	157	expert	no	0.97 [0.91, 1.00]	1.00 [0.95, 1.00]	Sensitivity
							Very serious^a	serious^b	Not applicable	No serious imprecision	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	No serious imprecision	VERY LOW
6 lead ECG with prototype recorder placed on thorax/abdomen in supine 1 measure of 5 seconds	1 Caldwell, 2012²⁹	157	expert	no	0.97 [0.91, 1.00]	1.00 [0.95, 1.00]	Sensitivity
							Very serious^a	serious^b	Not applicable	No serious imprecision	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	No serious imprecision	VERY LOW
6 lead ECG with standard electrode positions 1 measure of 5 seconds	1 Caldwell, 2012²⁹	157	expert	no	0.97 [0.91, 1.00]	1.00 [0.95, 1.00]	Sensitivity
							Very serious^a	serious^b	Not applicable	No serious imprecision	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	No serious imprecision	VERY LOW
6 lead ECG using Rhythm Pad device 1 measure of 10s	1 Sabar, 2019²²²	752	Auto Expert	No no	0.95 [0.87, 0.99] 0.94 [0.85, 0.98] Median^g: 0.94 [0.85, 0.98]	0.99 [0.97, 1.00] 0.97 [0.95, 0.98] 0.97 [0.95, 0.98]	Sensitivity
							Very serious^a	No serious indirectness	No serious inconsistency	serious^d	VERY LOW
							Specificity
							Very serious^a	No serious indirectness	No serious inconsistency	No serious imprecision	LOW

(a): Risk of bias was assessed using the QUADAS-2 checklist. The evidence was downgraded by 1 increment if the majority of studies were rated at high risk of bias, and downgraded by 2 increments if the majority of studies were rated at very high risk of bias.

(b): Indirectness was assessed using the QUADAS-2 checklist items referring to applicability. The evidence was downgraded by 1 increment if the majority of studies were seriously indirect, and downgraded by 2 increments if the majority of studies are very seriously indirect

(c): Inconsistency was assessed by visual inspection of the sensitivity/specificity plots, or data (if 2 studies). The evidence was downgraded by 1 increment if there was no overlap of 95% confidence intervals. For single studies no evaluation was made and ‘not applicable’ was recorded.

(d): Imprecision was assessed based on inspection of the confidence region in the diagnostic meta-analysis or, where diagnostic meta-analysis has not been conducted, assessed according to the range of confidence intervals in the individual studies. The evidence was downgraded by 1 increment when the confidence interval around the point estimate crossed one of the clinical thresholds (0.90 or 0.60), and downgraded by 2 increments when the confidence interval around the point estimate crossed both of the clinical thresholds. The upper clinical threshold marked the point above which recommendations would be possible, and the lower clinical threshold marked the point below which the tool would be regarded as of little clinical use.

(e): Pooled sensitivity/specificity from diagnostic meta-analysis

(f): indicates that because the raw data were not provided by the paper, and were also not able to be calculated by the reviewer, these were not able to be included in any diagnostic meta-analyses

(g): For non-pooled analyses, the median sensitivity and the paired specificity were reported. If there were an even number of studies the lower middle value was given.

Table 10Clinical evidence summary: diagnostic test accuracy for other non-12 lead ECG (12 lead ECG interpreted by expert cardiologist/electrophysiologist as gold standard)

Where 95% CIs are provided in round brackets (or no 95% CIs are given), raw data were not available and Forest Plots or pooled analyses were not possible.

Index Test	Number of studies	n	Interpreter of index test	Gold standard simultaneous with index test?	Sensitivity (95% CI)	Specificity (95% CI)	Risk of bias	Indirectness	Inconsistency	Imprecision	GRADE
Limb lead ECG 1 measure of unknown duration	1 Mant, 2007¹⁶¹ Mant, 2007¹⁶¹	1484	Clinician (GP) Clinician (nurse)	Yes yes	0.83 [0.75, 0.89] 0.72 [0.63, 0.80] Median^g: 0.72 [0.63, 0.80]	0.89 [0.87, 0.90] 0.83 [0.81, 0.85] 0.83 [0.81, 0.85]	Sensitivity
							No serious risk of bias	No serious indirectness	No serious inconsistency	No serious imprecision	HIGH
							Specificity
							No serious risk of bias	No serious indirectness	serious^c	No serious imprecision	MOD
Chest lead ECG 1 measure of unknown duration	1 Mant, 2007¹⁶¹ Mant, 2007¹⁶¹	1484	Clinician (GP) Clinician (nurse)	Yes yes	0.85 [0.78, 0.90] 0.69 [0.60, 0.76] Median^g: 0.69 [0.60, 0.76]	0.86 [0.84, 0.88] 0.98 [0.97, 0.99] 0.98 [0.97, 0.99]	Sensitivity
							No serious risk of bias	No serious indirectness	serious^c	No serious imprecision	MOD
							Specificity
							No serious risk of bias	No serious indirectness	serious^c	No serious imprecision	MOD
Computer interpretation of V1, V4 leads information only using Philips resting 12 lead algorithm 1 measure of 10s	1 Greg, 2008⁸²	1785	automated	yes	0.88 [0.80, 0.93]	0.99 [0.98, 0.99]	Sensitivity
							Very serious^a	serious^b	Not applicable	serious^d	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	No serious imprecision	VERY LOW
Computer interpretation of V2, V5 leads information only using Philips resting 12 lead algorithm 1 measure of 10s	1 Greg, 2008⁸²	1785	automated	yes	0.84 [0.76, 0.91]	0.99 [0.98, 0.99]	Sensitivity
							Very serious^a	serious^b	Not applicable	serious^d	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	No serious imprecision	VERY LOW
bipolar lead 1 ECG (no other details) 1 measure of unknown duration	1 Somerville, 2000²⁴⁰	86	Clinician (GP) Expert (nurse)	No no	0.96 [0.80, 1.00] 0.92 [0.75, 0.99] Median^g: 0.92 [0.75, 0.99]	0.98 [0.91, 1.00] 0.88 [0.77, 0.95] 0.88 [0.77, 0.95]	Sensitivity
							Very serious^a	serious^b	serious^c	serious^d	VERY LOW
							Specificity
							Very serious^a	serious^b	serious^c	serious^d	VERY LOW

(a): Risk of bias was assessed using the QUADAS-2 checklist. The evidence was downgraded by 1 increment if the majority of studies were rated at high risk of bias, and downgraded by 2 increments if the majority of studies were rated at very high risk of bias.

(b): Indirectness was assessed using the QUADAS-2 checklist items referring to applicability. The evidence was downgraded by 1 increment if the majority of studies were seriously indirect, and downgraded by 2 increments if the majority of studies are very seriously indirect

(c): Inconsistency was assessed by visual inspection of the sensitivity/specificity plots, or data (if 2 studies). The evidence was downgraded by 1 increment if there was no overlap of 95% confidence intervals. For single studies no evaluation was made and ‘not applicable’ was recorded.

(d): Imprecision was assessed based on inspection of the confidence region in the diagnostic meta-analysis or, where diagnostic meta-analysis has not been conducted, assessed according to the range of confidence intervals in the individual studies. The evidence was downgraded by 1 increment when the confidence interval around the point estimate crossed one of the clinical thresholds (0.90 or 0.60), and downgraded by 2 increments when the confidence interval around the point estimate crossed both of the clinical thresholds. The upper clinical threshold marked the point above which recommendations would be possible, and the lower clinical threshold marked the point below which the tool would be regarded as of little clinical use.

(e): Pooled sensitivity/specificity from diagnostic meta-analysis

(f): indicates that because the raw data were not provided by the paper, and were also not able to be calculated by the reviewer, these were not able to be included in any diagnostic meta-analyses

(g): For non-pooled analyses, the median sensitivity and the paired specificity were reported. If there were an even number of studies the lower middle value was given.

Table 11Clinical evidence summary: diagnostic test accuracy for 12 lead ECG interpreted by automated algorithm or non-expert interpreters (12 lead ECG interpreted by expert cardiologist/electrophysiologist as gold standard)

Where 95% CIs are provided in round brackets (or no 95% CIs are given), raw data were not available and Forest Plots or pooled analyses were not possible.

Index Test	Number of studies	n	Interpreter of index test	Gold standard simultaneous with index	Sensitivity (95% CI)	Specificity (95% CI)	Risk of bias	Indirectness	Inconsistency	Imprecsion	GRADE
12 lead ECG interpreted by non-expert interpreter (GP or nurse 1 measure of unknown duration	5 Hald, 2017⁹¹ Kvist, 2019¹⁴⁰ Mant, 2007¹⁶¹ Mant, 2007¹⁶¹ Rhys, 2013²¹⁰ Somerville, 2000²⁴⁰ Somerville, 2000²⁴⁰	2999	Clinician (GP/nurse) Clinician (nurse) Clinician (GP) Clinician (nurse Clinician (GP) Clinician (GP) Clinician (nurse)	yes no Yes Yes yes No No	1.00 [0.69, 1.00] 0.97 [0.90, 1.00] 0.78 [0.69, 0.86] 0.77 [0.67, 0.85] 1.00 [0.16, 1.00] 1.00 [0.87, 1.00] 1.00 [0.75, 1.00] Pooled:0.89(0.77-0.96)^e	0.96 [0.89, 0.99] 1.00 [1.00, 1.00] 0.92 [0.90, 0.93] 0.85 [0.83, 0.87] 1.00 [0.88, 1.00] 0.98 [0.91, 1.00] 0.76 [0.55, 0.91] Pooled:0.97(0.85-0.99)^e	Sensitivity
							serious^a	serious^b	serious^c	serious^d	VERY LOW
							Specificity
							serious^a	serious^b	serious^c	serious^d	VERY LOW
12 lead ECG interpreted by non-expert interpreter (GP or nurse) combined with interpretive algorithm 1 measure of unknown duration	1 Mant, 2007¹⁶¹	1454	Clinician + automated		0.92 [0.85, 0.96]	0.91 [0.89, 0.93]	Sensitivity
							No serious risk of bias	No serious indirectness	Not applicable	serious^d	MOD
							Specificity
							No serious risk of bias	No serious indirectness	Not applicable	serious^d	MOD
12 lead ECG detection algorithm based on a co-efficient of variation of the beat intervals (CV). Threshold set at 0.12 1 measure of 10s	1 Langley, 2012¹⁴⁵	2124	automated	yes	0.905^f	0.896^f	Sensitivity
							No serious risk of bias	serious^b	Not applicable	Not assessed	MOD
							Specificity
							No serious risk of bias	serious^b	Not applicable	Not assessed	MOD
12 lead ECG detection algorithm based on the co-efficient of sample entropy (COSEn). Threshold set at −1.19 1 measure of 10s	1 Langley, 2012¹⁴⁵	2124	automated	yes	0.952^f	0.934^f	Sensitivity
							No serious risk of bias	serious^b	Not applicable	Not assessed	MOD
							Specificity
							No serious risk of bias	serious^b	Not applicable	Not assessed	MOD
12 lead ECG detection algorithm based on the mean successive beat interval difference (defined as the mean absolute successive beat interval difference divided by the mean beat interval (Delta). Threshold set at 0.11 1 measure of 10s	1 Langley, 2012¹⁴⁵	2124	automated	yes	0.905^f	0.893^f	Sensitivity
							No serious risk of bias	serious^b	Not applicable	Not assessed	MOD
							Specificity
							No serious risk of bias	serious^b	Not applicable	Not assessed	MOD
12 lead ECG interpreted by algorithm in Cardioview interpretive software (not 1described) measure of unknown duration	1 Rhys, 2013²¹⁰	32	automated	yes	1.00 [0.16, 1.00]	1.00 [0.88, 1.00]	Sensitivity
							Very serious^a	serious^b	Not applicable	Very serious^d	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	serious^d	VERY LOW
12 lead ECG interpreted by computer-based rhythm diagnosis (GE Healthcare Technologies MUSE software 005C, version 19) (automated) 1 measure of unknown duration	1 Poon, 2005¹⁹⁵	3954	automated	yes	0.91 [0.87, 0.94]	0.99 [0.99, 0.99]	Sensitivity
							serious^a	serious^b	Not applicable	serious^d	VERY LOW
							Specificity
							serious^a	serious^b	Not applicable	No serious imprecision	LOW
12 lead interpretive software (no details given) 1 measure of unknown duration	1 Mant, 2007¹⁶¹	2556	automated	yes	0.83 [0.78, 0.88]	0.99 [0.99, 0.99]	Sensitivity
							No serious risk of bias	No serious indirectness	Not applicable	No serious imprecision	HIGH
							Specificity
							No serious risk of bias	No serious indirectness	Not applicable	No serious imprecision	HIGH
Algorithm for reading 12 lead ECGs. This first tested for the presence of non-coupled P waves. If non-coupled P waves were detected the rhythm was considered non-atrial fibrillation and no further testing was done. If the rhythm did not have noncoupled P waves, and the percent power in each lead II or V1 was >=32% the rhythm was considered AF. This algorithm was derived from the ‘training set’ of 72 rhythms in the developmental analysis. 1 measure of unknown duration	1 Slocum, 1992²³⁷	82	automated	yes	0.68 [0.52, 0.82]	0.88 [0.74, 0.96]	Sensitivity
							serious^a	serious^b	Not applicable	serious^d	VERY LOW
							Specificity
							serious^a	serious^b	Not applicable	serious^d	VERY LOW
Computer interpretation of full 12 lead ECG V1-V6 using Philips resting 12 lead algorithm 1 measure of 10s	1 Greg, 2008⁸²	1785	automated	yes	0.89 [0.82, 0.94]	0.99 [0.98, 0.99]	Sensitivity
							Very serious^a	serious^b	Not applicable	serious^d	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	No serious imprecision	VERY LOW

(a): Risk of bias was assessed using the QUADAS-2 checklist. The evidence was downgraded by 1 increment if the majority of studies were rated at high risk of bias, and downgraded by 2 increments if the majority of studies were rated at very high risk of bias.

(b): Indirectness was assessed using the QUADAS-2 checklist items referring to applicability. The evidence was downgraded by 1 increment if the majority of studies were seriously indirect, and downgraded by 2 increments if the majority of studies are very seriously indirect

(c): Inconsistency was assessed by visual inspection of the sensitivity/specificity plots, or data (if 2 studies). The evidence was downgraded by 1 increment if there was no overlap of 95% confidence intervals. For single studies no evaluation was made and ‘not applicable’ was recorded.

(d): Imprecision was assessed based on inspection of the confidence region in the diagnostic meta-analysis or, where diagnostic meta-analysis has not been conducted, assessed according to the range of confidence intervals in the individual studies. The evidence was downgraded by 1 increment when the confidence interval around the point estimate crossed one of the clinical thresholds (0.90 or 0.60), and downgraded by 2 increments when the confidence interval around the point estimate crossed both of the clinical thresholds. The upper clinical threshold marked the point above which recommendations would be possible, and the lower clinical threshold marked the point below which the tool would be regarded as of little clinical use.

(e): Pooled sensitivity/specificity from diagnostic meta-analysis

(f): indicates that because the raw data were not provided by the paper, and were also not able to be calculated by the reviewer, these were not able to be included in any diagnostic meta-analyses

(g): For non-pooled analyses, the median sensitivity and the paired specificity were reported. If there were an even number of studies the lower middle value was given.

Table 12Clinical evidence summary: diagnostic test accuracy for blood pressure monitors (>24 hour ambulatory monitoring as gold standard)

Where 95% CIs are provided in round brackets (or no 95% CIs are given), raw data were not available and Forest Plots or pooled analyses were not possible.

Index Test	Number of studies	n	Interpreter of index test	Gold standard simultaneous with index test?	Sensitivity (95% CI)	Specificity (95% CI)	Risk of bias	Indirectness	Inconsistency	Imprecsion	GRADE
24 hr ambulatory Microlife Afib device Watch BP 1 reading every 20 mins, over 24 hrs	1 Kollias, 2018¹³²	5778	automated	yes	0.93 [0.91, 0.94]	0.98 [0.98, 0.99]	Sensitivity
							Serious^a	No serious indirectness	Not applicable	no serious imprecision	MOD
							Specificity
							Serious^a	No serious indirectness	Not applicable	no serious imprecision	MOD
AF-BP monitor device, to take home and use daily for 30 days If event detected had to take 2 additional readings. If 2 or3 indicated AF took a final reading 1 hr later	1 Wiesel, 2013²⁷⁸	139	automated	yes	1.00 [0.77, 1.00]	0.90 [0.83, 0.94]	Sensitivity
							Serious^a	No serious indirectness	Not applicable	serious^d	LOW
							Specificity
							Serious^a	No serious indirectness	Not applicable	serious^d	LOW

(a): Risk of bias was assessed using the QUADAS-2 checklist. The evidence was downgraded by 1 increment if the majority of studies were rated at high risk of bias, and downgraded by 2 increments if the majority of studies were rated at very high risk of bias.

(b): Indirectness was assessed using the QUADAS-2 checklist items referring to applicability. The evidence was downgraded by 1 increment if the majority of studies were seriously indirect, and downgraded by 2 increments if the majority of studies are very seriously indirect

(c): Inconsistency was assessed by visual inspection of the sensitivity/specificity plots, or data (if 2 studies). The evidence was downgraded by 1 increment if there was no overlap of 95% confidence intervals. For single studies no evaluation was made and ‘not applicable’ was recorded.

(d): Imprecision was assessed based on inspection of the confidence region in the diagnostic meta-analysis or, where diagnostic meta-analysis has not been conducted, assessed according to the range of confidence intervals in the individual studies. The evidence was downgraded by 1 increment when the confidence interval around the point estimate crossed one of the clinical thresholds (0.90 or 0.60), and downgraded by 2 increments when the confidence interval around the point estimate crossed both of the clinical thresholds. The upper clinical threshold marked the point above which recommendations would be possible, and the lower clinical threshold marked the point below which the tool would be regarded as of little clinical use.

(e): Pooled sensitivity/specificity from diagnostic meta-analysis

(f): indicates that because the raw data were not provided by the paper, and were also not able to be calculated by the reviewer, these were not able to be included in any diagnostic meta-analyses

(g): For non-pooled analyses, the median sensitivity and the paired specificity were reported. If there were an even number of studies the lower middle value was given.

Table 13Clinical evidence summary: diagnostic test accuracy for <7 day Holter devices (7 day Holter as gold standard)

Where 95% CIs are provided in round brackets (or no 95% CIs are given), raw data were not available and Forest Plots or pooled analyses were not possible.

Index Test	Number of studies	n	Interpreter of index test	Gold standard simultaneous with index test?	Sensitivity (95% CI)	Specificity (95% CI)	Risk of bias	Indirectness	Inconsistency	Imprecision	GRADE
1 day Holter	1 Mulder, 2012¹⁷¹	96	automated	yes	0.52 [0.30, 0.74]^**	1.00 [0.95, 1.00]^**	Sensitivity
							serious^a	serious^b	Not applicable	serious^d	VERY LOW
							Sensitivity
							serious^a	serious^b	Not applicable	No serious imprecision	LOW
2 day Holter	1 Mulder, 2012¹⁷¹	96	automated	yes	0.67 [0.43, 0.85]^**	1.00 [0.95, 1.00]^**	Sensitivity
							serious^a	serious^b	Not applicable	serious^d	VERY LOW
							Sensitivity
							serious^a	serious^b	Not applicable	No serious imprecision	LOW
3 day Holter	1 Mulder, 2012¹⁷¹	96	automated	yes	0.81 [0.58, 0.95]^**	1.00 [0.95, 1.00]^**	Sensitivity
							serious^a	serious^b	Not applicable	Very serious^d	VERY LOW
							Sensitivity
							serious^a	serious^b	Not applicable	No serious imprecision	LOW
4 day Holter	1 Mulder, 2012¹⁷¹	96	automated	yes	0.86 [0.64, 0.97]^**	1.00 [0.95, 1.00]^**	Sensitivity
							serious^a	serious^b	Not applicable	serious^d	VERY LOW
							Sensitivity
							serious^a	serious^b	Not applicable	No serious imprecision	LOW
5 day Holter	1 Mulder, 2012¹⁷¹	96	automated	yes	0.90 [0.70, 0.99]^**	1.00 [0.95, 1.00]^**	Sensitivity
							serious^a	serious^b	Not applicable	serious^d	VERY LOW
							Sensitivity
							serious^a	serious^b	Not applicable	No serious imprecision	LOW
6 day Holter	1 Mulder, 2012¹⁷¹	96	automated	yes	0.95 [0.76, 1.00]^**	1.00 [0.95, 1.00]^**	Sensitivity
							serious^a	serious^b	Not applicable	serious^d	VERY LOW
							Specificity
							serious^a	serious^b	Not applicable	No serious imprecision	LOW

(a): Risk of bias was assessed using the QUADAS-2 checklist. The evidence was downgraded by 1 increment if the majority of studies were rated at high risk of bias, and downgraded by 2 increments if the majority of studies were rated at very high risk of bias.

(b): Indirectness was assessed using the QUADAS-2 checklist items referring to applicability. The evidence was downgraded by 1 increment if the majority of studies were seriously indirect, and downgraded by 2 increments if the majority of studies are very seriously indirect

(c): Inconsistency was assessed by visual inspection of the sensitivity/specificity plots, or data (if 2 studies). The evidence was downgraded by 1 increment if there was no overlap of 95% confidence intervals. For single studies no evaluation was made and ‘not applicable’ was recorded.

(d): Imprecision was assessed based on inspection of the confidence region in the diagnostic meta-analysis or, where diagnostic meta-analysis has not been conducted, assessed according to the range of confidence intervals in the individual studies. The evidence was downgraded by 1 increment when the confidence interval around the point estimate crossed one of the clinical thresholds (0.90 or 0.60), and downgraded by 2 increments when the confidence interval around the point estimate crossed both of the clinical thresholds. The upper clinical threshold marked the point above which recommendations would be possible, and the lower clinical threshold marked the point below which the tool would be regarded as of little clinical use.

(e): Pooled sensitivity/specificity from diagnostic meta-analysis

(f): indicates that because the raw data were not provided by the paper, and were also not able to be calculated by the reviewer, these were not able to be included in any diagnostic meta-analyses

(g): For non-pooled analyses, the median sensitivity and the paired specificity were reported. If there were an even number of studies the lower middle value was given.

**: unit of analysis was each of the separate measures done on each person

Table 14Clinical evidence summary: other longer term devices (>24 hour ambulatory monitoring as gold standard)

Where 95% CIs are provided in round brackets (or no 95% CIs are given), raw data were not available and Forest Plots or pooled analyses were not possible.

Index Test	Number of studies	n	Interpreter of index test	Gold standard simultaneous with index test?	Sensitivity (95% CI)	Specificity (95% CI)	Risk of bias	Indirectness	Inconsistency	Imprecision	GRADE
R.Test Evolution 3 – triggered ECG	1 Roten, 2012²¹⁹	100	expert	yes	0.88 [0.74, 0.96]	1.00 [0.94, 1.00]	Sensitivity
							Very serious^a	serious^b	Not applicable	serious^d	VERY LOW
							Specificity
							Very serious^a	serious^b	Not applicable	No serious imprecision	VERY LOW
R.Test Evolution 4 (NorDiaTech, Paris, France) - External loop recorder (ELR) 1 measure × 48 hours	1 Sejr, 2019²³³ Sejr, 2019²³³	1412	Automated expert	Yes yes	0.92 [0.79, 0.98] 0.84 [0.69, 0.94] Median^g: 0.84 [0.69, 0.94]	0.87 [0.85, 0.89] 0.98 [0.97, 0.99] 0.98 [0.97, 0.99]	Sensitivity
							No serious risk of bias	No serious indirectness	No serious inconsistency	serious^d	MOD
							Specificity
							No serious risk of bias	No serious indirectness	serious^c	No serious imprecision
Vitaphone 3100 BT external loop recorder. Recorded event ECGs manually when triggered by the patient or automatically 1 measure × 24 hours	2 Muller, 2009¹⁷² Velthuis, 2013²⁶⁸	104	Automated automated	Yes yes	1.00 [0.86, 1.00] 0.95 [0.86, 0.99]^ Median^g: 0.95 [0.86, 0.99]**	0.50 [0.29, 0.71] 0.51 [0.49, 0.53]^ 0.51 [0.49, 0.53]**	Sensitivity
							serious^a	serious^b	No serious inconsistency	serious^d	VERY LOW
							Specificity
							serious^a	serious^b	No serious inconsistency	No serious imprecision	LOW
SRAclinic, Apoplex Medical Technologies. Stroke Risk Analysis (SRA) – software analysis of every hourly ECG snippet of continuous (non 12 lead) ECG monitoring, and report sent daily to stroke unit.(automated) threshold of 0-1=SR and 2 or more =AF 1 measure of >24 hours	1 Ross, 2018²¹⁸	798	automated	yes	0.98(0.952-0.990)^** No raw data in paper – CIs provided by paper)	0.27(0.223-0.322)^** No raw data in paper – CIs provided by paper)	Sensitivity
							serious^a	No serious indirectness	Not applicable	No serious imprecision	MOD
							Specificity
							serious^a	No serious indirectness	Not applicable	No serious imprecision	MOD
SRAclinic, Apoplex medical Technologies. Stroke Risk Analysis (SRA) – software analysis of every hourly ECG snippet of continuous (non 12 lead) ECG monitoring, and report sent daily to stroke unit.(automated) threshold of 0-2=SR and 3 or more =AF 1 measure of >24 hours	1 Ross, 2018²¹⁸	798	automated	yes	0.840(0.790-0.878)^** No raw data in paper – CIs provided by paper)	0.700(0.645-0.749)^** No raw data in paper – CIs provided by paper)	Sensitivity
							serious^a	No serious indirectness	Not applicable	No serious imprecision	MOD
							Specificity
							serious^a	No serious indirectness	Not applicable	No serious imprecision	MOD
Standard ECG monitoring devices without AF-RS - with ECG confirmation if positive (thus specificity was 100%) 1 measure of 48 hours	1 Arevalo-Manso, 2016⁷	59	Automated/clinician	No	0.08 [0.00, 0.36]	1.00 [0.92, 1.00]	Sensitivity
							Very serious^a	No serious indirectness	Not applicable	No serious imprecision	LOW
							Specificity
							Very serious^a	No serious indirectness	Not applicable	No serious imprecision	LOW
4 lead AF-RS monitor - with ECG confirmation if positive (thus specificity was 100%) 1 measure of 48 hours	1 Arevalo-Manso, 2016⁷	17	Automated/clinician	No	0.57 [0.18, 0.90]	1.00 [0.69, 1.00]	Sensitivity
							Very serious^a	No serious indirectness	Not applicable	Very serious^d	LOW
							Specificity
							Very serious^a	No serious indirectness	Not applicable	serious^d	LOW
12-bit resolution digital ECG recoding for 1-2 hours 1 measure of 1-2 hours	1 Rizos, 2010²¹⁴	136	automated	yes	0.72 [0.53, 0.87]	0.63 [0.53, 0.72]	Sensitivity
							serious^a	No serious indirectness	Not applicable	serious^d	LOW
							Specificity
							serious^a	No serious indirectness	Not applicable	serious^d	LOW
6 channel Holter (H12+, Mortara Instruments) 1 measure of 24 hours	1 Rizos, 2010²¹⁴	136	expert	yes	0.23 [0.05, 0.54]	1.00 [0.97, 1.00]	Sensitivity
							Very serious^a	No serious indirectness	Not applicable	No serious imprecision	LOW
							Specificity
							Very serious^a	No serious indirectness	Not applicable	No serious imprecision	LOW
Thumb ECG (Zenicor Medical Systems AB) 30s measure twice daily for 30 days	1 Poulsen, 2017¹⁹⁶	95	expert	yes	0.59 [0.33, 0.82]	0.87 [0.78, 0.94]	Sensitivity
							Very serious^a	No serious indirectness	Not applicable	serious^d	VERY LOW
							Specificity
							Very serious^a	No serious indirectness	Not applicable	serious^d	VERY LOW
Kardia-Band Mean 11.3 hrs per day for 110 days	1 Wasserlauf, 2019²⁷⁵	26	automated	yes	0.977^f	0.989^f	Sensitivity
							Very serious^a	No serious indirectness	Not applicable	Not assessed	LOW
							Specificity
							Very serious^a	No serious indirectness	Not applicable	Not assessed	LOW
>24 hour telemetry with ‘electrocardiomatrix’ Median 46 hours	1 Brown, 2019²⁴	260	automated	yes	0.98 [0.95, 0.99]	0.86 [0.71, 0.95]	Sensitivity
							Very serious^a	No serious indirectness	Not applicable	No serious imprecision	LOW
							Specificity
							Very serious^a	No serious indirectness	Not applicable	serious^d	VERY LOW
>WiPatch – 2 lead skinpatch ECG system 24 hours	1 Karunadas, 2020¹²⁶	141	automated	yes	1.0 [0.29, 1.0]	1.0 [0.97, 0.95]	Sensitivity
							Serious^a	No serious indirectness	Not applicable	Very serious imprecision^d	VERY LOW
							Specificity
							Serious^a	No serious indirectness	Not applicable	No serious imprecision	MODERATE
One-off 12 lead ECG (not a long-term detection device but tested against the long-term gold-standard so has been included for completeness) Unclear but probably a standard 10s ECG	1 Lyckhage, 2020¹⁶⁰	366	unclear	no	0.18 [0.04, 0.43]	1.0 [0.99, 1.0]	Sensitivity
							Very serious^a	No serious indirectness	Not applicable	No serious imprecision	LOW
							Specificity
							Very serious^a	No serious indirectness	Not applicable	No serious imprecision	LOW
One-off pulse palpation (not a long-term detection approach but tested against the long-term gold-standard so has been included for completeness) Unclear duration	1 Lyckhage, 2020¹⁶⁰	366	unclear	no	0.47 [0.23, 0.72]	0.67 [0.62, 0.72]	Sensitivity
							Very serious^a	No serious indirectness	Not applicable	serious^d	VERY LOW
							Specificity
							Very serious^a	No serious indirectness	Not applicable	No serious imprecision	LOW
Single lead (MP1) patch-based ambulatory ECG monitor 24 hours duration*	1 Lai, 2020¹⁴⁴	40	automated	yes	0.931^f	0.934^f	Sensitivity
							serious^a	serious^b	Not applicable	Not asssessed	LOW
							Specificity
							serious^a	serious^b	Not applicable	Not asssessed	LOW

(a): Risk of bias was assessed using the QUADAS-2 checklist. The evidence was downgraded by 1 increment if the majority of studies were rated at high risk of bias, and downgraded by 2 increments if the majority of studies were rated at very high risk of bias.

(b): Indirectness was assessed using the QUADAS-2 checklist items referring to applicability. The evidence was downgraded by 1 increment if the majority of studies were seriously indirect, and downgraded by 2 increments if the majority of studies are very seriously indirect

(c): Inconsistency was assessed by visual inspection of the sensitivity/specificity plots, or data (if 2 studies). The evidence was downgraded by 1 increment if there was no overlap of 95% confidence intervals. For single studies no evaluation was made and ‘not applicable’ was recorded.

(d): Imprecision was assessed based on inspection of the confidence region in the diagnostic meta-analysis or, where diagnostic meta-analysis has not been conducted, assessed according to the range of confidence intervals in the individual studies. The evidence was downgraded by 1 increment when the confidence interval around the point estimate crossed one of the clinical thresholds (0.90 or 0.60), and downgraded by 2 increments when the confidence interval around the point estimate crossed both of the clinical thresholds. The upper clinical threshold marked the point above which recommendations would be possible, and the lower clinical threshold marked the point below which the tool would be regarded as of little clinical use.

(e): Pooled sensitivity/specificity from diagnostic meta-analysis

(f): indicates that because the raw data were not provided by the paper, and were also not able to be calculated by the reviewer, these were not able to be included in any diagnostic meta-analyses

(g): For non-pooled analyses, the median sensitivity and the paired specificity were reported. If there were an even number of studies the lower middle value was given.

**: unit of analysis was each of the separate measures done on each person

Accuracy of tests for detection

Authors

1. Detection diagnostic accuracy

1.1. Review question: What are the most accurate methods for detecting atrial fibrillation in people with cardiovascular risk factors for AF and/or symptoms suggestive of AF?

1.2. Introduction

1.3. PICO table

1.4. Methods and process

1.5. Clinical evidence

1.5.1. Included studies

1.5.2. Excluded studies

1.5.3. Summary of clinical studies included in the evidence review (Gold standard = 12 lead ECG stratum)

1.5.4.

1.5.5. Summary of clinical studies included in the evidence review (Gold standard = >24 hours ambulatory monitoring stratum)

1.5.6. Quality assessment of clinical studies included in the evidence review

STRATUM 1: 12 lead ECG interpreted by expert cardiologist/electrophysiologist as gold standard

1.5.7. Quality assessment of clinical studies included in the evidence review

STRATUM 2: >24 hour ambulatory monitoring [such as Holter] as gold standard

1.6. Economic evidence

1.7. The committee’s discussion of the evidence

1.7.1. Interpreting the evidence

1.7.1.1. The outcomes that matter most

1.7.1.2. The quality of the evidence

1.7.1.3. Benefits and harms

1.7.2. Cost effectiveness and resource use

1.7.3. Other factors the committee took into account

References

Appendices

Appendix A. Review protocols

Appendix B. Literature search strategies

Appendix C. Clinical evidence selection

Appendix D. Clinical evidence tables

Appendix E. Coupled sensitivity and specificity forest plots and sROC curves

Appendix F. Health economic evidence selection

Appendix G. Health economic evidence tables

Appendix H. QUADAS2 risk of bias assessment

Appendix I. Excluded studies

I.1. Excluded clinical studies

I.2. Excluded health economic studies

Appendix J. Research recommendations

Table 1PICO characteristics of review question

Table 2Summary of studies included in the evidence review for detection of atrial fibrillation

Table 3Summary of studies included in the evidence review for detection of atrial fibrillation

Table 4Clinical evidence summary: diagnostic test accuracy for mobile ECG devices (12 lead ECG interpreted by expert cardiologist/electrophysiologist as gold standard)

Table 5Clinical evidence summary: diagnostic test accuracy for blood pressure monitors (12 lead ECG interpreted by expert cardiologist/electrophysiologist as gold standard)

Table 6Clinical evidence summary: diagnostic test accuracy for pulse palpation (12 lead ECG interpreted by expert cardiologist/electrophysiologist as gold standard)

Table 7Clinical evidence summary: diagnostic test accuracy for photoplethysmography (12 lead ECG interpreted by expert cardiologist/electrophysiologist as gold standard)

Table 8Clinical evidence summary: diagnostic test accuracy for 3-lead tele ECG (12 lead ECG interpreted by expert cardiologist/electrophysiologist as gold standard)

Table 9Clinical evidence summary: diagnostic test accuracy for 6 lead ECG (12 lead ECG interpreted by expert cardiologist/electrophysiologist as gold standard)

Table 10Clinical evidence summary: diagnostic test accuracy for other non-12 lead ECG (12 lead ECG interpreted by expert cardiologist/electrophysiologist as gold standard)

Table 11Clinical evidence summary: diagnostic test accuracy for 12 lead ECG interpreted by automated algorithm or non-expert interpreters (12 lead ECG interpreted by expert cardiologist/electrophysiologist as gold standard)

Table 12Clinical evidence summary: diagnostic test accuracy for blood pressure monitors (>24 hour ambulatory monitoring as gold standard)

Table 13Clinical evidence summary: diagnostic test accuracy for <7 day Holter devices (7 day Holter as gold standard)

Table 14Clinical evidence summary: other longer term devices (>24 hour ambulatory monitoring as gold standard)