Chung-Wah Siu2, Chi-Chung Choy1, Chi-Kin Chan3 Effectiveness of community atrial fibrillation screening in over 10,000 citizens using smartphone electrocardiogram- The AFinder program Ngai-Yin Chan1, Chung-Wah Siu2, Chi-Chung Choy1, Chi-Kin Chan3 Department of Medicine & Geriatrics, Princess Margaret Hospital, Hong Kong Department of Medicine, Queen Mary Hospital, Hong Kong Department of Medicine, United Christian Hospital, Hong Kong

- I have nothing to declare

Funding The AFinder program was financially supported by Bayer. Bayer has no influence on the design, implementation, data analysis and preparation of report of this study.

Is AF Suitable for Screening ? It is an important health problem with an accepted treatment There are facilities for diagnosis and treatment There is a latent and symptomatic stage The natural history is understood There is an agreed policy on whom to treat The cost of finding is economically balanced with overall health The case finding is a continuous process The screening test should be suitable and acceptable to the population Wilson J, Junger G. Principles and practice of screening for disease. In: Public Health paper No. 34. Edn. Geneva: World Health Organization, 1968.

Current Recommendation for AF Screening Kirchhof P et al. 2016 ESC guidelines for the management of AF developed in collaboration with EACTS. EHJ 2016;37:2893-962.

AF Screening in the Pharmacy Setting Lowres N et al. Feasibility and cost- effectiveness of stroke prevention through community screening for atrial fibrillation using iPhone ECG in pharmacies. The SEARCH-AF study. Thromb Haemost 2014;111:1167-76.

Community AF Screening In a community AF screening study on 13,122 Hong Kong citizens, 0.8% detection rate (number-needed-to-screen=125) for newly diagnosed AF was observed.1 In STROKESTOP study, a detection rate of 3% (NNS=33) for newly diagnosed AF was observed with intermittent ECG recordings over 2 weeks in 7,173 citizens in Sweden.2 Chan NY et al. Screening for atrial fibrillation in 13,122 Hong Kong citizens with smartphone electrocardiogram. Heart 2017;103:24-31. Svennberg E et al. Mass screening for untreated atrial fibrillation. Circulation 2015;131:2176-84.

AF Screening in the Primary Care Setting Chan PH et al. Diagnostic performance of a smartphone-based photoplethysmographic application for atrial fibrillation screening in a primary care setting. JAHA 2016;5(7). pii:e003428

AF Screening With Automatic BP Measuring Device Sensitivity = 80.6% Specificity = 98.7% Positive predictive value = 42.4% Negative predictive value = 99.8% Chan PH 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.

AFinder Program – Objective and Outcome Measures We aim to examine the effectiveness of a non-governmental organization-led, community-based AF screening program carried out by trained layperson volunteers. The primary outcome measures are NNS for one newly diagnosed AF and NNS for one appropriately treated newly diagnosed AF. The secondary outcome measures are proportion of under-treated known AF and the diagnostic performance of the automated detection algorithm of the smartphone app in detecting AF.

AFinder Program – Methodology A territory-wide, community-based, systematic AF screening program Initiated by a panel of 4 cardiologists Developed, organized and led by a non-governmental organization, named the Hong Kong Council of Social Service (HKCSS) First medical program under the World Health Organization Network for Age- friendly Cities and Communities in Hong Kong

Engagement and Empowerment of Elderly Citizens A focus group discussion with elderly citizens was held to contribute to the design of the program 84 layperson volunteers aged over 50 were recruited and trained to use the Kardia mobile device to perform smartphone-based single- lead ECG for community AF screening

AFinder Program- A Resourceful Website

AFinder Program-Screened Population, Screening Activities and Procedures The screening program was publicized via different channels including media promotion and placement of posters in community centres by different NGO’s in Hong Kong All citizens aged 50 or above were eligible for participation 118 screening sessions in 108 community centres were conducted in the period between November 8 2015 to September 23 2016

AFinder Program-Screened Population, Screening Activities and Procedures For every participant, his/her sex and age were recorded and a single 30-second smartphone ECG was performed ECG’s were downloaded by a call centre weekly and sent to a cardiologist for review Citizens with AF were phone contacted for completing a baseline and 9-month FU questionnaire The ECG reports were mailed to those participants with AF and they were advised to seek medical attention

AFinder Program- Baseline and 9-month FU Questionnaires Baseline questionnaire : -Previous history of AF -Symptoms of AF -Medical conditions 9-month questionnaire : -Whether medical attention had been sought -Whether oral anticoagulant, Aspirin or Clopidogrel was prescribed -Level of drug compliance (very good=compliant all the time, fair=compliant > 50% of time, unsatisfactory=compliant <50% of time, poor=not taking the drug at all)

AFinder Program- Results 839 (7.2%) smartphone ECG’s were uninterpretable NNS for 1 newly diagnosed AF=145

AFinder Program- Results 25 (0.23%) participants did not seek medical attention 30 (0.28%) participants were not prescribed oral anticoagulation (17 were given Aspirin, 1 was given Clopidogrel and 12 were not given any antithrombotic drug) NNS for 1 newly diagnosed AF receiving appropriate oral anticoagulation=671

Results- Characteristics of Newly Diagnosed AF All participants with AF (n=244) Newly diagnosed AF (n=74) Known AF (n=133) p-value Age 79.5±7.9 81.1±7.3 78.1±8.1 0.007 Sex (F), n(%) 172 (70.5) 51 (68.9) 97 (72.9) 0.542 Medical conditions Heart failure, n(%) 17 (7.0) 6 (8.1) 9 (6.8) 0.738 Hypertension, n(%) 50 (67.6) 95 (71.4) 0.569 Diabetes, n(%) 63 (25.8) 19 (25.7) 34 (25.6) 0.973 Stroke, n(%) 40 (16.4) 9 (12.2) 22 (16.5) 0.0004 Coronary artery disease, n(%) 25 (10.2) 7 (9.5) 12 (9.0) 0.920 Peripheral artery disease, n(%) 8 (3.3) 7 (5.3) 0.045

Results- Characteristics of Newly Diagnosed AF All participants with AF (n=244) Newly diagnosed AF (n=74) Known AF (n=133) p-value CHA2D2VASc score 3.9±1.5 3.9±1.6 3.9±1.4 0.990 0 in men and 1 in women, n (%) 5 (2) 2 (2.7) 0.055 1 in men and 2 in women, n (%) 13 (5.3) 3 (4.1) 10 (7.5) 0.338 ≥2 in men and ≥3 in women, n (%) 226 (92.6) 69 (93.2) 123 (92.5) 0.823

Results- Characteristics of Newly Diagnosed AF All participants with AF (n=244) Newly diagnosed AF (n=74) Known AF (n=133) p-value Symptoms of AF Palpitations, n (%) 18 (24.3) Shortness of breath, n (%) 25 (33.8) Decrease in exercise tolerance, n(%) 19 (25.7) Dizziness, n(%) 16 (21.6) Syncope, n(%) 7 (9.5) Chest pain, n(%) 10 (13.5) Asymptomatic, n(%) 36 (48)

Treatment Rates for Newly Diagnosed and Known AF 29.8% 250 200 33.8% 150 22.2% 100 50 Newly diagnosed and Newly diagnosed AF known AF Known AF Appropriately treated with oral anticoaguation Not appropriately treated with oral anticoagulation

Results- Characteristics of Under-treated Known AF All known AF (n=133) Under-treated known AF (n=88) Treated known AF (n=45) p-value Age 78.1±8.1 78.2±7.6 77.8±9.1 0.8 Sex (F), n(%) 97 (72.9) 67 (76.1) 30 (66.7) 0.248 Medical conditions Heart failure, n(%) 9 (6.8) 7 (8.0) 2 (4.4) 0.464 Hypertension, n(%) 95 (71.4) 64 (72.7) 33 (73.3) 0.934 Diabetes, n(%) 34 (25.6) 22 (25) 12 (26.7) 0.834 Stroke, n(%) 22 (16.5) 8 (9.1) 14 (31.1) 0.001 Coronary artery disease, n(%) 12 (9.0) 5 (5.7) 7 (15.6) 0.064 Peripheral artery disease, n(%) 7 (5.3) 2 (2.3) 5 (11.1) 0.033

Results- Characteristics of Under-treated Known AF CHA2D2VASc score All known AF (n=133) 3.9±1.4 Under-treated known AF (n=88) 3.7±1.3 Treated known AF (n=45) 4.3±1.5 p-value 0.042 0 in men and 1 in women, n (%) 1 in men and 2 in women, n (%) 10 (7.5) 8 (9.1) 2 (4.4) 0.331 ≥2 in men and ≥3 in women, n (%) 123 (92.5) 80 (90.9) 43 (95.6)

Results- Characteristics of Under-treated Known AF Among 88 patients with under-treated known AF, 23 (26.1%; 95% CI: 16.9- 35.5; 22 on Aspirin and 1 on Clopidogrel) were on antiplatelet drugs

Results- Diagnostic Performance of Automated Detection Algorithm for AF Sensitivity = 75%; Specificity = 98.2% Positive predictive value = 64.9%; negative predictive value = 99.5%

Conclusions A new format and approach of AF screening in the community by an NGO-led program carried out by trained layperson volunteers is feasible and capable of detecting citizens with newly diagnosed AF with an NNS similar to other programs The effectiveness of the program in subsequently leading participants with newly diagnosed AF or under-treated known AF to receive appropriate oral anticoagulation therapy is weakened by the lack of a structured downstream management pathway The specificity and negative predictive value of the automated AF detection algorithm of the Kardia mobile device are excellent but both the sensitivity and positive predictive value are still suboptimal; the sensitivity of the algorithm has to be further improved before it can be considered as a tool for auto-screening

Limitations The participants self-reported their history of AF, other comorbid conditions, health -seeking behaviour, drug history and compliance to treatment The management outcomes in other domains like treatment of cardiovascular conditions, rate control and rhythm control were not asked during the 9-month follow-up survey Smartphone single-lead ECG was used for the diagnosis of AF, conventional 12- lead ECG was not performed as reference The uninterpretable rate of smartphone ECG’s was relatively high The cost-effectiveness of this program has not been evaluated Whether this new approach of community AF screening program is applicable to other regions is unknown

Acknowledgement We sincerely thank all the elderly citizens who contributed to the design of the AFinder screening program, all the workers who conducted the community screening for AF and all the participants of the program. Last but not the least, we thank HKCSS for leading and implementing this large-scale community program