Diagnostic accuracy and yield of screening tests for atrial fibrillation in the family practice setting: a multicentre cohort study

Abstract

Background: Detection of undiagnosed or undertreated ("actionable") atrial fibrillation could increase the use of appropriate oral anticoagulant therapy and reduce the risk of stroke. We sought to compare newer screening technologies with a pulse-check for the detection of atrial fibrillation and to determine whether the detection of actionable atrial fibrillation increases the use of oral anticoagulant agents.

Methods: This prospective multicentre cohort study involved 22 primary care clinics. We recruited participants aged 65 years and older who were attending routine appointments. Each participant underwent 3 methods of screening: a 30-second radial pulse-check; single-lead electrocardiogram; and screening by blood pressure machine with atrial fibrillation detection algorithms. Participants who received a positive result on 1 or more test underwent 12-lead electrocardiogram with or withour 24-hour Holter. Screening tests were compared using the McNemar test. Participants with confirmed atrial fibrillation received follow-up at 90 days.

Results: The mean age of participants was 73.7 (± 6.9) years, and 53.4% of participants were female. Of 2171 patients, we had data from all 3 screening tests for 2054 patients. Both single-lead electrocardiogram and the blood pressure device showed superior specificity compared with pulse-check (p < 0.001 for each). Fifty-six patients (2.7%) had confirmed atrial fibrillation: 12 patients had newly detected atrial fibrillation (none of the patients were using anticoagulation agents), and 44 patients had previously diagnosed atrial fibrillation (42 patients were receiving anticoagulant therapy, 2 were not). Thus, 14 patients had actionable atrial fibrillation (0.7%). By 90 days, 77% of patients with actionable atrial fibrillation had started anticoagulant therapy.

Interpretation: Newer screening technologies showed superior specificity compared with a pulse-check. Screening detected undiagnosed or undertreated atrial fibrillation in 0.7% of participants, and 77% started appropriate anticoagulant therapy.

Trial registration: ClinicalTrials.gov, no. NCT02262351.

Conflict of interest statement

Competing interests: Russell Quinn reports consulting fees from Boehringer Ingelheim, Servier and Bayer, and has research grants from Boehringer Ingelheim and Bayer. Richard Birtwhistle reports research funding from Eli Lilly, Shire Pharmaceuticals, Merck and Pfizer. Jeffrey Healey reports consulting fees from Bayer, Boehringer Ingelheim, Pfizer, Bristol-Myers Squib, and has research grants from Boehringer Ingelheim, Bayer, Pfizer and Bristol-Myers Squibb. No other competing interests were declared.

Copyright 2018, Joule Inc. or its licensors.

Figures

Figure 1:

Flow of patients through the study. Percentages in parentheses relate to numbers in preceding level, with the exception of those marked by an asterisk (*), which indicates that the percentage was calculated based on the 2054 patients with data from all 3 screening tests. Note: AF = atrial fibrillation, AFL = atrial flutter, ECG = electrocardiogram, ICD = implantable cardioverter defibrillator, OAC = oral anticoagulant agent, SL-ECG = single-lead electrocardiogram.