Research ArticleOriginal ResearchA

Diagnostic Accuracy of a Smartphone-Operated, Single-Lead Electrocardiography Device for Detection of Rhythm and Conduction Abnormalities in Primary Care

Jelle C.L. Himmelreich, Evert P.M. Karregat, Wim A.M. Lucassen, Henk C.P.M. van Weert, Joris R. de Groot, M. Louis Handoko, Robin Nijveldt and Ralf E. Harskamp
The Annals of Family Medicine September 2019, 17 (5) 403-411; DOI: https://doi.org/10.1370/afm.2438

Article Figures & Data

Figures

Tables

  • Table 1

    Baseline Characteristics of the Study Population

    VariableOverall (n = 214)Patients With Symptom-Driven ECG (n = 108)Patients With Protocol-Driven ECG (n = 106)
    Demographics
    Age, y64.1 ± 14.759.1 ± 16.369.3 ± 10.7a
    Male115 (53.7)55 (50.9)44 (41.5)
    History
    Obesity (BMI >30 kg/m2)41 (19.2)12 (11.1)29 (27.4)a
    Smoking
     Current smoker36 (16.8)17 (15.7)19 (17.9)
     Past history of smoking72 (33.6)23 (21.3)49 (46.2)a
     No history of smoking72 (33.6)42 (38.9)30 (28.3)
     Unknown34 (15.9)26 (24.1)8 (7.5)a
    Alcohol abuse10 (4.7)5 (4.6)5 (4.7)
    Hypertension87 (40.7)31 (28.7)56 (52.8)a
    Diabetes66 (30.8)10 (9.3)56 (52.8)a
    Hypercholesterolemia54 (25.2)20 (18.5)34 (32.1)a
    Atrial fibrillation or flutter23 (10.7)13 (12.0)10 (9.4)
    Other arrhythmia12 (5.6)6 (5.6)6 (5.7)
    Coronary heart disease21 (9.8)4 (3.7)17 (16.0)a
    TIA or ischemic stroke13 (6.1)6 (5.6)7 (6.6)
    Valvular heart disease9 (4.2)6 (5.6)3 (2.8)
    Heart failure8 (3.7)5 (4.6)3 (2.8)
    Chronic obstructive pulmonary disease21 (9.8)9 (8.3)12 (11.3)
    Peripheral vascular disease19 (8.9)4 (3.7)15 (14.2)a
    Chronic renal failure26 (12.1)9 (8.3)17 (16.0)
     eGFR of those with chronic renal failure (mL/min/1.73m2)50.3 ± 6.048.9 ± 8.651.0 ± 4.5
    Medication
    Beta blocker42 (19.6)15 (13.9)27 (25.5)a
    Calcium-channel blocker31 (14.5)11 (10.2)20 (18.9)
    Digoxin1 (0.5)0 (0)1 (0.9)
    Potassium-channel blocker2 (0.9)2 (1.9)0 (0)
    Sodium-channel blocker1 (0.5)1 (0.9)0 (0)

    • BMI = body mass index; ECG = electrocardiography; eGFR = estimated glomerular filtration rate; TIA = transient ischemic attack.

    • Note: Data are presented as mean ± SD for continuous variables and no. (%) for categorical variables.

    • ↵a P <.05.

  • Table 2

    Indications for Undergoing 12L-ECG (n = 214)

    IndicationNo. (%)
    Symptom-driven ECG (n = 108)
     Palpitations48 (44.4)
     Other chest symptoms (nonacute)47 (43.5)
     Dyspnea23 (21.3)
     Lightheadedness16 (14.8)
     Fatigue14 (13.0)
     Collapse3 (2.8)
     Other17 (15.7)
    Protocol-driven ECG (n = 106)
     Cardiovascular risk management34 (32.1)
     Known diabetes mellitus45 (42.5)
     Known ischemic heart disease13 (12.3)
     Known heart rhythm disorder7 (6.6)
     Known TIA or ischemic stroke4 (3.8)
     Known heart failure1 (0.9)
     Irregular pulse at examination1 (0.9)
     Follow-up after starting new medication1 (0.9)

    • 12L = 12-lead; ECG = electrocardiography; TIA = transient ischemic attack.

    • Note: We included patients once and reported 1 reason for ECG per patient. See Supplemental Appendix, http://www.AnnFamMed.org/content/17/5/403/suppl/DC1/, for how we handled patients with >1 symptom and/or comorbidity.

  • Table 3

    Outcomes of 12L-ECG (n = 214)

    OutcomeNo. (%)
    Rhythm
     Sinus rhythm187 (87.4)
     Atrial fibrillation20 (9.3)
     Atrial flutter3 (1.4)
     Narrow complex tachycardia3 (1.4)
     Broad complex tachycardia0 (0)
     Ectopic atrial rhythm1 (0.5)
    Ectopic beats
     Premature atrial complex7 (3.3)
     Premature ventricular complex16 (7.5)
    Conduction abnormalities
     AV block7 (3.3)
      First-degree AV block7 (100)
      Second-degree AV block, Wenckebach0 (0)
      Second-degree AV block, Mobitz II0 (0)
      Third-degree AV block0 (0)
     Bundle branch block23 (10.7)
      LBBB5 (21.7)
      RBBB9 (39.1)
      LAD/LAFB9 (39.1)
    Composite outcomes
     Atrial fibrillation or flutter23 (10.7)
     Any rhythm abnormalitya44 (20.6)
     Any conduction abnormalityb28 (13.1)
  • Table 4

    Diagnostic Accuracy Measures of the Interpretation of 1L-ECG by Cardiologists or the Smartphone Algorithm Using 12L-ECG as Reference Standard

    Outcome Assessed% Sensitivity (95% CI)% Specificity (95% CI)LR+ (95% CI)LR– (95% CI)% PPV (95% CI)% NPV (95% CI)
    Atrial fibrillation or flutter
     Cardiologists100 (85.2-100)100 (98.1-100)∞a0b100c100c
     Smartphone algorithm87.0 (66.4-97.2)97.9 (94.7-99.4)41.5 (15.5-110.9)0.13 (0.05-0.38)83.3 (65.2-93.0)98.4 (95.6-99.4)
    Any rhythm abnormality
     Cardiologists90.9 (78.3-97.5)93.5 (88.7-96.7)14.1 (7.9-25.1)0.10 (0.04-0.25)78.4 (67.1-86.7)97.6 (94.0-99.0)
    Any conduction abnormality
     Cardiologists46.4 (27.5-66.1)100 (98.0-100)∞a0.54 (0.38-0.76)100c92.5 (89.8-94.6)

    • 1L = single-lead; 12L = 12-lead; ECG = electrocardiography; LR+ = positive likelihood ratio; LR– = negative likelihood ratio; NPV = negative predictive value; PPV = positive predictive value.

    • ↵a LR+ is infinite and 95% CI is not applicable when specificity = 100%.14

    • ↵b LR– is 0 and 95% CI is not applicable when sensitivity = 100%.14

    • ↵c 95% CI is not applicable when PPV or NPV = 100%.14,15

Additional Files

  • Supplemental Appendixes

    Supplemental appendixes

    Files in this Data Supplement:

  • The Article in Brief

    Diagnostic Accuracy of a Smartphone-Operated, Single-Lead Electrocardiography Device for Detection of Rhythm and Conduction Abnormalities in Primary Care

    Jelle C.L. Himmelreich , and colleagues

    Background Performing a traditional electrocardiogram in primary care can be cumbersome and is unavailable in some primary care practices. A new alternative--a handheld, smartphone-enabled 1-lead ECG--may provide physicians with a viable alternative for detecting cardiac arrhythmias in patients presenting with non-acute cardiac concerns in primary care.

    What This Study Found This validation study tested the diagnostic accuracy of a smartphone-operated, single-lead ECG (1L-ECG, AliveCor KardiaMobile) with integrated algorithm for atrial fibrillation (AF) against 12-lead ECG in a primary care population. In a multi-center validation study of the device's diagnostic accuracy, blinded cardiologists assessed data from 214 Dutch primary care patients collected simultaneously from the 1L and 12L devices. The handheld device showed excellent diagnostic accuracy, as cardiologists were able to detect all cases of atrial fibrillation and atrial flutter from the device's readings. The study also compared expert review of the handheld device's readings to the smartphone-integrated diagnostic algorithm's interpretation of the device's output. The algorithm correctly identified 87% of atrial fibrillation cases and 98% of non-atrial fibrillation cases. The algorithm was less accurate in categorizing other abnormalities.

    Implications

    • The findings of this study suggest that the smartphone-operated ECG may be a viable alternative for physicians in need of a point-of-care device to detect key arrhythmias in the preventive care of stroke.

  • Annals Journal Club

    Sep/Oct 2019 Annals Journal Club


    Divya Manda and Michael E. Johansen, Associate Editor

    The Annals of Family Medicine encourages readers to develop a learning community to improve health care and health through enhanced primary care. Participate by conducting a RADICAL journal club. RADICAL stands for Read, Ask, Discuss, Inquire, Collaborate, Act, and Learn. We encourage diverse participants to think critically about important issues affecting primary care and act on those discussions.1

    HOW IT WORKS

    In each issue, the Annals selects an article and provides discussion tips and questions. Take a RADICAL approach to these materials and post a summary of your conversation in our online discussion. (Open the article and click on "TRACK Discussion/ Submit a comment.") Discussion questions and information are online at: http://www.AnnFamMed.org/site/AJC/.

    CURRENT SELECTION

    Article for Discussion

    Himmelreich JC, Karregat EP, Lucassen WA, et al. Diagnostic accuracy of a smartphone-operated, single-lead ECG for detection of rhythm and conduction abnormalities in primary care. Ann Fam Med. 2019;17(5):403-411.

    Discussion Tips

    Many patients present to primary care physicians with symptoms that are suspicious for cardiac arrhythmias, but 12-lead ECGs to evaluate the symptoms are not always available or are cumbersome. This article describes a blinded case series to evaluate the utility of a smartphone-enabled 1-lead ECG.

    Discussion Questions

    • What question is asked by this study and why does it matter?
    • How is this study different from previous studies about smartphone-enabled ECGs? What does this study add to the field?
    • How strong is the study design for answering the question?
    • Define sensitivity, specificity, likelihood ratios, positive/negative predictive value. How are these calculated?
      • What are some of criticisms about the use of these values? 23
      • How the main variables were measured?
    • What were the primary and secondary findings of the study? How accurate was 1-lead ECG for detecting atrial fibrillation/atrial flutter vs other arrhythmias vs ectopic beats? Is there value in combining these end points?
    • Importantly, what does this study not investigate?
    • How patients were selected, excluded, or lost to follow-up; how the main variables were measured?; confounding variables; and how the findings were interpreted?
  • To what degree can the findings be accounted for by:
    • How patients were selected, excluded, or lost to follow-up; how the main variables were measured?; confounding variables; and how the findings were interpreted?
  • How applicable are the study results to your patient population? how applicable are the study results in a typical primary care office? What is the transportability of the findings?
  • How might spectrum bias be relevant to this study?
  • How might the study change your practice? Would you be more likely to use a smartphone-enabled 1-lead ECG in your office?
  • What are the limitations of the study and how may this limit the applicability of the results?
  • How does this study relate to and differ from the discussion around using smartphones ECG as screening tool?4
  • What are the next steps in applying the findings to clinical practice and in primary care?
  • What questions remain regarding the use of smartphone-enabled 1-lead ECGs in primary care?

    • References

    1. Stange KC, Miller WL, McLellan LA, et al. Annals Journal Club: It's time to get RADICAL. Ann Fam Med. 2006;4(3):196-197.
    2. Janssens AC, Deng Y, Borsboom GJ, Eijkemans MJ, Habbema JD, Steyerberg EW. A new logistic regression approach for the evaluation of diagnostic test results. Med Decis Making. 2005; 25(2): 168-177.
    3. McNutt R, Hadler NM. Rethinking our thinking about diagnostic tests: there is nothing positive or negative about a test result. https://thehealthcareblog.com/blog/2015/12/01/rethinking-about-diagnostic-tests-there-is-nothing-positive-or-negative-about-a-test-result/.Accessed Aug 9, 2019.
    4. Mandrola J, Foy A. Downsides of detecting atrial fibrillation in asymptomatic patients. Am Fam Physician. 2019; 99(6): 354-355.

Back to top

In this issue

Print
Download PDF
Article Alerts
Email Article
Citation Tools
Get Permissions

Jump to section

Subjects

Keywords