Screening for Atrial Fibrillation With Electrocardiography: Evidence Report and Systematic Review for the US Preventive Services Task Force

Importance  Atrial fibrillation is the most common arrhythmia and increases the risk of stroke.

Objective  To review the evidence on screening for nonvalvular atrial fibrillation with electrocardiography (ECG) and stroke prevention treatment in asymptomatic adults 65 years or older to inform the US Preventive Services Task Force.

Data Sources  MEDLINE, Cochrane Library, and trial registries through May 2017; references; experts; literature surveillance through June 6, 2018.

Study Selection  English-language randomized clinical trials (RCTs), prospective cohort studies evaluating detection rates of atrial fibrillation or harms of screening, and systematic reviews evaluating stroke prevention treatment. Eligible treatment studies compared warfarin, aspirin, or novel oral anticoagulants (NOACs) with placebo or no treatment. Studies were excluded that focused on persons with a history of cardiovascular disease.

Data Extraction and Synthesis  Dual review of abstracts, full-text articles, and study quality. When at least 3 similar studies were available, random-effects meta-analyses were conducted.

Main Outcomes and Measures  Detection of previously undiagnosed atrial fibrillation, mortality, stroke, stroke-related morbidity, and harms.

Results  Seventeen studies were included (n = 135 300). No studies evaluated screening compared with no screening and focused on health outcomes. Systematic screening with ECG identified more new cases of atrial fibrillation than no screening (absolute increase, from 0.6% [95% CI, 0.1%-0.9%] to 2.8% [95% CI, 0.9%-4.7%] over 12 months; 2 RCTs, n = 15 803), but a systematic approach using ECG did not detect more cases than an approach using pulse palpation (2 RCTs, n = 17 803). For potential harms, no eligible studies compared screening with no screening. Warfarin (mean, 1.5 years) was associated with a reduced risk of ischemic stroke (relative risk [RR], 0.32 [95% CI, 0.20-0.51]) and all-cause mortality (RR, 0.68 [95% CI, 0.50-0.93]) and with increased risk of bleeding (5 trials, n = 2415). Participants in treatment trials were not screen detected, and most had long-standing persistent atrial fibrillation. A network meta-analysis reported that NOACs were associated with a significantly lower risk of a composite outcome of stroke and systemic embolism (adjusted odds ratios compared with placebo or control ranged from 0.32-0.44); the risk of bleeding was increased (adjusted odds ratios, 1.4-2.2), but confidence intervals were wide and differences between groups were not statistically significant.

Conclusions and Relevance  Although screening with ECG can detect previously unknown cases of atrial fibrillation, it has not been shown to detect more cases than screening focused on pulse palpation. Treatments for atrial fibrillation reduce the risk of stroke and all-cause mortality and increase the risk of bleeding, but trials have not assessed whether treatment of screen-detected asymptomatic older adults results in better health outcomes than treatment after detection by usual care or after symptoms develop.

Atrial fibrillation is a supraventricular tachyarrhythmia characterized by uncoordinated electrical activity and resulting inefficient atrial contraction.¹ Atrial fibrillation can be categorized as paroxysmal, persistent, permanent, and nonvalvular (eTable 1 in the Supplement). Prevalence increases with age, from less than 0.2% for persons younger than 55 years to about 10% for those 85 years or older, and is higher for men than women (eTable 2 in the Supplement).² About 25% of atrial fibrillation is paroxysmal.³

Atrial fibrillation increases the risk for stroke and thromboembolism by reducing cardiac blood flow and predisposing to thrombus formation, particularly in the left atrial appendage.^1,4 For persons with atrial fibrillation, the annual incidence of stroke increases by about 1.5% per decade, from 1.3% in persons aged 50 to 59 years to 5.1% in persons aged 80 to 89 years.⁵ Strokes attributable to atrial fibrillation are associated with a poor prognosis.^6-9 Approximately 30% of patients with atrial fibrillation who have a stroke die within 1 year of the stroke, and up to 30% of survivors are permanently disabled.¹⁰

Patients may not notice symptoms of atrial fibrillation before a serious event, such as stroke. Of patients who have a stroke attributable to atrial fibrillation, an estimated 20% or more are diagnosed with atrial fibrillation at the time of the stroke or shortly thereafter.^11-13 Thus, identifying asymptomatic atrial fibrillation and starting anticoagulation therapy may prevent strokes and deaths. To inform a recommendation by the US Preventive Services Task Force (USPSTF), the evidence on detection of previously undiagnosed atrial fibrillation and benefits and harms of screening and stroke prevention treatment for atrial fibrillation in populations and settings relevant to US primary care was reviewed.

Detailed methods and additional details of results and analyses are available in the full evidence report at https://www.uspreventiveservicestaskforce.org/Page/Document/UpdateSummaryFinal/atrial-fibrillation-screening-with-electrocardiography. Additional descriptions of studies evaluating aspirin and other published systematic reviews are available in the full evidence report. Figure 1 shows the analytic framework and key questions (KQs) that guided the review.

PubMed/MEDLINE and the Cochrane Library were searched for English-language articles published from inception through May 2017. Search strategies are listed in the eMethods in the Supplement. ClinicalTrials.gov and the World Health Organization International Clinical Trials Registry platform were searched for unpublished literature. To supplement electronic searches, investigators reviewed reference lists of pertinent articles, studies suggested by reviewers, and comments received during public commenting periods. Since May 2017, ongoing surveillance was conducted through article alerts and targeted searches of journals to identify major studies published in the interim that may affect the conclusions or understanding of the evidence and the related USPSTF recommendation. The last surveillance was conducted on June 6, 2018 and identified no eligible studies.

Two investigators independently reviewed titles, abstracts, and full-text articles to determine eligibility using prespecified criteria for each key question (eTable 3 in the Supplement). Disagreements were resolved by discussion. The review included English-language studies focused on adults 65 years or older conducted in countries categorized as “very high” on the United Nations Human Development Index. Only studies rated as good or fair quality were included. Studies focused on adults with a history of stroke, transient ischemic attack (TIA), coronary heart disease, or heart failure were excluded.

For KQ1 (direct evidence that screening improves health outcomes), KQ2 (detection of undiagnosed atrial fibrillation), and KQ3 (harms of screening), studies were required to enroll unselected or explicitly asymptomatic adults. Eligible screening tests included systematic screening with electrocardiography (ECG) (eg, 12-lead ECG, intermittent use of handheld ECG) or systematic screening with both pulse palpation and ECG for all participants. Studies whose interventions used other technologies (eg, blood pressure monitoring) or physical examination only were excluded. For KQ1, randomized clinical trials (RCTs) and nonrandomized controlled intervention studies were eligible. For KQ2 and KQ3, prospective cohort studies were also eligible.

For benefits (KQ4) and harms (KQ5) of stroke prevention treatment, eligible studies compared treatment with aspirin or oral anticoagulants (warfarin or the novel oral anticoagulants [NOACs] apixaban, dabigatran, edoxaban, or rivaroxaban) vs placebo or no treatment controls. RCTs and nonrandomized controlled intervention studies were eligible. Systematic reviews of trials were also eligible if they were directly relevant (eg, focused on primary prevention; included the relevant warfarin trials). For KQ5, prospective cohort studies were also eligible.

For each included study, 1 investigator extracted pertinent information about the populations, tests or treatments, comparators, outcomes, settings, and designs, and a second investigator reviewed for completeness and accuracy. Two independent investigators assessed the quality of studies as good, fair, or poor, using predefined criteria developed by the USPSTF and adapted for this topic.¹⁴ Disagreements were resolved by discussion. Individual study quality ratings are reported in eTables 4-9 in the Supplement.

Findings for each question were summarized in tabular and narrative format. To determine whether meta-analyses were appropriate, clinical and methodological heterogeneity were assessed. When at least 3 similar studies were available, quantitative synthesis was conducted with random-effects models using the inverse-variance weighted method (DerSimonian and Laird) to estimate pooled effects.¹⁵ For KQ2, fewer than 3 studies were available for each comparison, and absolute risk differences and odds ratios (ORs) were calculated for detection of unknown atrial fibrillation. For KQ4 and KQ5, relative risks (RRs) and 95% CIs were calculated for all-cause mortality, cardiovascular-related mortality, ischemic stroke, moderately to severely disabling stroke, TIA, major bleeding, major extracranial bleeding, intracerebral hemorrhage, minor bleeding, and a composite outcome of ischemic stroke or intracerebral hemorrhage.

For all quantitative syntheses, the I² statistic was calculated to assess statistical heterogeneity.^16,17 Quantitative analyses were conducted using Comprehensive Meta-Analysis version 3.3 (Biostat Inc) and Stata version 14 (StataCorp).

The overall strength of the body of evidence was assessed for each key question as high, moderate, low, or insufficient using methods developed for the USPSTF (and the Evidence-based Practice Center program), based on the overall quality of studies, consistency of results between studies, precision of findings, and risk of reporting bias.¹⁴

A total of 17 studies (22 articles) with 135 300 participants were included (Figure 2). The main results for each key question are summarized below.

Key Question 1. Does screening for atrial fibrillation with ECG improve health outcomes (ie, reduce all-cause mortality or reduce morbidity or mortality from stroke) in asymptomatic older adults?

Key Question 1a. Does improvement in health outcomes vary for subgroups defined by stroke risk (eg, based on CHA₂DS₂-VASc score), age, sex, or race/ethnicity?

No eligible studies were identified that focused on this question and reported results. One fair-quality RCT of 1001 participants and a primary outcome of time to diagnosis of atrial fibrillation, the Assessment of Remote Heart Rhythm Sampling Using the AliveCor Heart Monitor to Screen for Atrial Fibrillation (REHEARSE-AF) trial (described in KQ2), reported limited information on health outcomes but was not designed or powered to evaluate them.¹⁹ For all-cause mortality, the authors reported 3 deaths in the screening group and 5 in the no screening group (P = .51). For a composite of stroke, TIA, or systemic embolism, there were 6 vs 10 events, respectively (hazard ratio [HR], 0.6 [95% CI, 0.2-1.7]; P = .34).

Another RCT, the STROKESTOP study, is ongoing (anticipated completion, March 2019) and has not yet reported results for the outcomes and comparisons eligible for this review.^20-22 The primary outcome is incidence of stroke over 5 years. The study randomized 28 768 persons aged 75 to 76 years in 2 regions in Sweden to screening program invitations for atrial fibrillation or no invitations. The screening program used an initial 12-lead ECG and then a handheld 1-lead ECG recorder for intermittent recordings over 2 weeks. The detection rate for previously unknown atrial fibrillation was 3.0% (95% CI, 2.7%-3.5%; n = 218/7173) in the intervention group; incidence data for atrial fibrillation in the control group have not yet been reported. Of the new cases detected in the intervention group, few were identified on the initial 12-lead ECG (37/218 [17%]). More than 90% of patients with newly diagnosed atrial fibrillation accepted initiation of oral anticoagulant therapy.

Key Question 2. Does systematic screening for atrial fibrillation with ECG identify older adults with previously undiagnosed atrial fibrillation more effectively than usual care?

The review included 3 fair-quality RCTs described in 7 articles (Table 1).^19,23-28 All 3 trials were conducted in the United Kingdom. Two compared systematic screening (with pulse palpation and single-lead or 12-lead ECG) with opportunistic screening^23-28; 1 of those also included a comparison with no screening.^23-27 One trial compared systematic, twice-weekly screening using a single-lead handheld ECG vs no screening.¹⁹

The Screening for Atrial Fibrillation in the Elderly (SAFE) study was a multicenter cluster trial (n = 14 802) that randomized 50 practices to screening vs no screening.^23-27 Within the 25 practices randomized to screening, individual participants were randomized to systematic screening or opportunistic screening. For the screening practices, primary care physicians and other members of the health care team attended educational days covering the importance of detecting atrial fibrillation and available treatment options. Another trial randomized 3001 participants from 4 practices.²⁸ Nurses conducting screenings received 2 hours of training on the assessment of the pulse rhythm. The third trial, REHEARSE-AF, randomized 1001 participants to a twice-weekly screening with a single-lead ECG using a handheld device vs no screening.¹⁹ All trials enrolled patients 65 years or older; the mean age of participants was 72 to 75 years. Only REHEARSE-AF reported baseline stroke risk scores for participants; the mean CHA₂DS₂-VASc (congestive heart failure, hypertension, age ≥75 years [doubled], diabetes, stroke/transient ischemic attack/thromboembolism [doubled], vascular disease [prior myocardial infarction, peripheral artery disease, or aortic plaque], age 65-74 years, sex category [female]) score was 3 (SD, 1). The SAFE study reported the CHADS₂ (congestive heart failure, hypertension, age ≥75 years, diabetes mellitus, prior stroke or transient ischemic attack or thromboembolism [doubled]) scores for the 149 newly identified cases of atrial fibrillation and reported that more cases in the systematic screening group had scores of 2 or more than in the opportunistic group, but the difference was not statistically significant (43.2% vs 29.3%, P = .08).²³

The trials did not find a statistically significant difference between systematic and opportunistic screening for detection of new cases of atrial fibrillation (Figure 3). The SAFE study found that more new cases of atrial fibrillation were detected in patients undergoing any screening (systematic or opportunistic) compared with no screening (149 vs 47; risk difference, 0.6% [95% CI, 0.2%-0.9%]; OR, 1.6 [95% CI, 1.1-2.2]). The subgroup analyses reported from the SAFE study found that screening may not increase detection of new cases among women. Men in the systematic (OR, 2.7 [95% CI, 1.5-4.7]) and opportunistic (OR, 2.3 [95% CI, 1.3-4.2]) screening groups had greater odds of having atrial fibrillation diagnosed than men in the no screening group. The odds were not significantly increased for women in either screening group compared with no screening (OR, 1.0 [95% CI, 0.6-1.6] and OR, 1.2 [95% CI, 0.7-1.9], respectively). Patients aged 65 to 74 years and those older than 75 years had similar odds of having atrial fibrillation diagnosed in both the systematic screening (OR, 1.6 [95% CI, 0.9-2.9] and OR, 1.6 [95% CI, 0.98-2.5], respectively) and opportunistic (OR, 1.6 [95% CI, 0.9-2.9] and OR, 1.6 [95% CI, 1.0-2.6], respectively) screening groups, compared with no screening. The REHEARSE-AF study reported that more new cases of atrial fibrillation were detected in those undergoing screening compared with no screening (19 vs 5; HR, 3.9 [95% CI, 1.4-10.4]; risk difference, 2.8% [95% CI, 0.9%-4.7%]; OR, 3.9 [95% CI, 1.5-10.6]).

Key Question 3. What are the harms of screening for atrial fibrillation with ECG in older adults?

Key Question 3a. Do the harms of screening vary for subgroups defined by stroke risk, age, sex, or race/ethnicity?

No eligible studies assessing labeling or harms of subsequent interventions initiated because of screening with ECG (eg, subsequent ablation with complications) were identified. One of the trials included for KQ2, the SAFE study, assessed anxiety associated with screening and provided limited evidence showing that anxiety scores were not significantly different for systematic and opportunistic screening groups after screening or after 17 months.^23,24 It did not, however, collect anxiety data from patients in the no screening group, which would have allowed for a comparison between screening and no screening.

Potential harms of screening include misinterpretation of ECGs and subsequent unnecessary treatment (eg, warfarin for someone without atrial fibrillation). An analysis of 2595 participants in the SAFE study from 49 general practices assessed the accuracy of general practitioners and interpretive software for diagnosing atrial fibrillation.²⁵ General practitioners missed 20 of 99 atrial fibrillation cases (20%) on 12-lead ECG and misinterpreted 114 of 1355 cases (8%) of normal sinus rhythm as atrial fibrillation, compared with reference standard cardiologists (sensitivity, 79.8% [95% CI, 70.5%-87.2%]; specificity, 91.6% [95% CI, 90.1%-93.1%]). False-positive rates varied from 0% to 44% for individual general practitioners (SD, 13%). Combining general practitioners’ interpretations with those of interpretive software increased the sensitivity (91.9% [95% CI, 86.6%-97.3%]), but specificity was about the same (91.1% [95% CI, 89.6%-92.6%]). Use of single-lead or limb-lead ECGs resulted in slightly lower specificity.

Key Question 4. What are the benefits of anticoagulation or antiplatelet therapy on health outcomes in asymptomatic, screen-detected older adults with atrial fibrillation?

Key Question 4a. Do the benefits of anticoagulation or antiplatelet therapy vary for subgroups defined by stroke risk, age, sex, or race/ethnicity?

No trials or systematic reviews that focused on asymptomatic, screen-detected participants were found. Six RCTs of persons who were not screen detected (Table 2) were included; most had long-standing persistent nonvalvular atrial fibrillation; prevalence of baseline or past symptoms (eg, palpitations, dyspnea) was generally not reported. Three RCTs evaluated warfarin,^30,33,34 1 evaluated aspirin,³⁵ and 2 (described in 3 articles) evaluated both warfarin and aspirin.^29,31,32 Seven systematic reviews (eTable 10 in the Supplement) were included: 3 were traditional systematic reviews with meta-analyses,^36-38 3 were meta-analyses of individual patient data,^39-41 and 1 was a network meta-analysis.⁴²

Five trials (6 articles) evaluated warfarin.^29-34 Four of the 5 trials compared warfarin with placebo (Atrial Fibrillation, Aspirin, and Antikoagulation study [AFASAK I],²⁹ Canadian Atrial Fibrillation Anticoagulation [CAFA],³³ Stroke Prevention in Atrial Fibrillation [SPAF I],^31,32 Stroke Prevention in Nonrheumatic Atrial Fibrillation [SPINAF]³⁴), and 1 (Boston Area Anticoagulation Trial for Atrial Fibrillation [BAATAF])³⁰ compared warfarin with no treatment. The BAATAF trial allowed participants in the no treatment group to take aspirin, but use of aspirin or other antithrombotic medications was not permitted in the 4 placebo-controlled trials. Two trials (AFASAK I and SPAF I) were 3-group studies that included aspirin groups (in addition to warfarin and placebo or no treatment). Two trials were double-blinded (CAFA, SPINAF), and 3 were open label (AFASAK I, BAATAF, SPAF I). All trials began in the 1980s, were completed by 1992, and were stopped early because of evidence favoring warfarin.

The mean age of participants in trials evaluating warfarin ranged from 67 to 74 years. Four trials enrolled fewer than 30% women and just 1 reported race or ethnicity (16% of participants were nonwhite in SPAF I). The baseline prevalence of hypertension and diabetes ranged from 32% to 58% and 12% to 18%, respectively. AFASAK I and SPINAF did not include participants with paroxysmal atrial fibrillation; the other 3 trials reported that 7% to 34% had paroxysmal atrial fibrillation. Most participants in the trials had atrial fibrillation for more than 1 year, although AFASAK I did not report information about the duration of atrial fibrillation before enrollment. Baseline stroke risk was not reported by the trials because stroke risk scores used in current practice were not yet developed. All trials titrated doses of warfarin using either prothrombin time or international normalized ratio (INR).

Warfarin treatment over an average of 1.5 years was associated with reductions in all-cause mortality (RR, 0.68 [95% CI, 0.50-0.93]), ischemic stroke (RR, 0.32 [95% CI, 0.20-0.51]), and moderately to severely disabling stroke (RR, 0.38 [95% CI, 0.19-0.78]; 5 trials, 2415 participants), compared with controls (Figure 4). For a population with baseline annual stroke risk of 4%, such as patients with CHADS₂ scores of 2, the results indicate that warfarin was associated with a number needed to treat of 24 (95% CI, 17-36) to prevent 1 ischemic stroke over 1.5 years. For a population of 1000 adults 65 years or older with an annual stroke risk of 4%, this translates to an absolute reduction of 28 ischemic strokes per year and an absolute reduction of 16 deaths per year.

Results of previously published systematic reviews were consistent with the findings of this review and provide some additional details about subgroups and about NOACs. An individual patient data meta-analysis reported that warfarin was associated with a reduction in stroke for both men and women, without a statistically significant difference between them (relative risk reduction, 60% [95% CI, 35%-76%] for men and 84% [95% CI, 55%-95%] for women).³⁹ Another individual patient data meta-analysis found that warfarin was associated with a reduced risk of ischemic stroke for all ages; for the assessment of relative benefit with increasing age, the interaction did not reach statistical significance (eg, HR, 0.22 [95% CI, 0.11-0.41] for 50-year-olds; HR, 0.53 [95% CI, 0.35-0.81] for 90-year-olds; P = .07 for age × warfarin interaction).⁴¹ A previously published network meta-analysis that used 21 RCTs (96 017 participants) of treatment for nonvalvular atrial fibrillation found that vitamin K antagonists and all 4 NOACs were associated with lower risk of a primary composite outcome of stroke (any type) and systemic embolism (eTable 10 in the Supplement).⁴² For the NOACs, the authors reported an association with a significant reduction in the primary outcome compared with placebo or control (unadjusted ORs from 0.27-0.38; adjusted ORs from 0.32-0.44) but no statistically significant differences for the NOACs in comparison with one another. In adjusted analyses (for CHADS₂ scores, time in therapeutic range, duration of follow-up), the NOACs were not statistically different from vitamin K antagonists (eTable 10 in the Supplement).

Key Question 5. What are the harms of anticoagulation or antiplatelet therapy in asymptomatic, screen-detected older adults with atrial fibrillation?

Key Question 5a. Do the harms of anticoagulation or antiplatelet therapy vary for subgroups defined by stroke risk, age, sex, or race/ethnicity?

All 6 RCTs described in KQ4 for benefits of warfarin or aspirin also reported harms (Table 2).^29-35 Seven systematic reviews (eTable 10 in the Supplement) were also included: 4 were traditional systematic reviews with meta-analyses,^36-38,43 2 were individual patient data meta-analyses,^39,41 and 1 was a network meta-analysis.⁴²

Warfarin treatment was associated with increased risk of major bleeding (RR, 1.8 [95% CI, 0.9-3.7]; 5 trials, 2415 participants), major extracranial bleeding (RR, 1.6 [95% CI, 0.7-3.6]; 4 trials, 1744 participants), and intracranial hemorrhage (RR, 1.9 [95% CI, 0.6-6.7]; 5 trials, 2415 participants) compared with controls, but confidence intervals were wide and differences between groups were not statistically significant (Figure 5). Across trials evaluating warfarin, 31 major bleeding events occurred, 20 in warfarin groups and 11 in control groups. Minor bleeding was much more common, with 136 events in warfarin groups and 86 in control groups over a mean of 1.6 years (pooled RR, 1.6 [95% CI, 1.2-2.0]; 4 trials, 1744 participants) (eFigure 1 in the Supplement).

Results of previously published systematic reviews were consistent with the findings of this review. The previously published network meta-analysis (96 017 participants) reported that the 4 NOACs were associated with an increased risk of major bleeding, but the confidence intervals were wide and differences between groups were not statistically significant (adjusted ORs from 1.4 to 2.2) (eTable 10 in the Supplement), and there were no statistically significant differences between any of the 4 NOACs.⁴² Compared with vitamin K antagonists, 3 of the NOACs (apixaban, dabigatran, and edoxaban) were associated with a lower risk of bleeding (range of ORs from 0.64 [95% CI, 0.46-0.90] to 0.85 [95% CI, 0.65-1.1]), but the difference was only statistically significant for edoxaban (OR, 0.64 [95% CI, 0.46-0.90]).

Table 3 provides the summary of findings. No eligible studies evaluated screening for atrial fibrillation with ECG compared with no screening and focused on health outcomes. One ongoing RCT, STROKESTOP, is designed to assess health outcomes for this comparison.^20,22 For harms of screening, no eligible studies provided information that allowed comparison between screening and no screening.

This review found that both 1-time systematic screening with 12-lead ECG and twice-weekly screening with a single-lead ECG identify more new cases of atrial fibrillation than no screening (absolute increase over 12 months, 0.6% [95% CI, 0.1%-0.9%] and 2.8% [95% CI, 0.9%-4.7%], respectively), but screening with ECG did not detect more cases than opportunistic screening with pulse palpation. REHEARSE-AF, STROKESTOP, and some uncontrolled studies^20,45,46 suggest that the number of new atrial fibrillation cases detected is greater with intermittent ECG recordings or continuous ECG recordings over 2 weeks than with a 1-time ECG.

Most asymptomatic older adults with previously unrecognized or undiagnosed atrial fibrillation have a stroke risk above the threshold for initiating anticoagulation (eContextual Question 2 in the Supplement). In STROKESTOP, more than 90% of patients with newly diagnosed atrial fibrillation were offered and accepted initiation of oral anticoagulant therapy.²⁰ The SAFE study reported that 78% of new cases identified with systematic screening had CHADS₂ scores of 1 or more and that 43% had scores of 2 or more.²³ If screening programs were implemented, they could be limited to persons 65 years or older who have CHA₂DS₂-VASc scores of 2 or higher to avoid screening persons for whom anticoagulation would not be indicated.

Potential harms of screening with ECG include misinterpretation of ECGs and subsequent treatments for persons without atrial fibrillation. Evidence suggests that some primary care physicians cannot accurately detect atrial fibrillation on an ECG.²⁵ For example, an analysis of 2595 participants from 49 general practices in central England participating in the SAFE study assessed the accuracy of general practitioners and interpretive software for diagnosing atrial fibrillation.²⁵ General practitioners misinterpreted (with or without the help of interpretive software) 8% of sinus rhythm cases as atrial fibrillation. The analysis did not evaluate the accuracy of primary care physicians for other ECG findings (eg, those suggesting ischemia) that might also lead to subsequent testing and interventions (eg, angiography) that could result in complications. Another study using a database from a US hospital that evaluated 2298 ECGs (from 1085 patients) with a computerized interpretation of atrial fibrillation found that ECGs from 382 patients (35%) had been misinterpreted; physicians did not correct the computerized misinterpretation and initiated inappropriate and potentially harmful treatments, and they pursued unnecessary additional testing for 92 patients (8.5%).⁴⁷

This review found consistent evidence that anticoagulation reduces the risk of stroke and all-cause mortality and increases the risk of bleeding for persons with nonvalvular atrial fibrillation who do not have a history of stroke or TIA. For a population of 1000 adults 65 years or older with an annual stroke risk of 4%, the results translate to an absolute reduction of 28 ischemic strokes per year, an absolute reduction of 16 deaths per year, and an absolute increase of 5 major bleeding events per year. A previously published network meta-analysis⁴² included in this review found that NOACs were not statistically different from vitamin K antagonists for a composite outcome (any stroke and systemic embolism) or for all-cause mortality.

All 5 included trials that evaluated warfarin began in the 1980s and were completed by 1992, and stroke incidence may have decreased since then with the increased use of statins and antihypertensive medications. Thus, the absolute benefits of anticoagulation might be less in the current era, although the relative benefits are likely similar. The current clinical approach to anticoagulation has evolved since the trials were conducted. There is some uncertainty about the INR target ranges of 3 trials (BAATAF, SPAF I, and SPINAF) because these trials used prothrombin time targets, and conversion of prothrombin time to INR cannot be precisely achieved because of uncertain sensitivity of thromboplastin agents. In addition, trials followed protocols (eg, for warfarin dosing); routine clinical practice may not be as rigorous. However, observational studies of anticoagulation suggest that results are similar in routine clinical practice.⁴⁸ In addition, the trials that evaluated warfarin had a mean duration of follow-up from 1.2 to 2.2 years (mean, 1.5 years) and were stopped early; thus, estimates for stroke and mortality reduction may not be applicable to lifelong anticoagulation. Although the review aimed to determine the benefits of treatment for asymptomatic, screen-detected older adults with nonvalvular atrial fibrillation, no trials or systematic reviews that focused on this population were found, and it is uncertain whether benefits of anticoagulation vary between symptomatic persons and asymptomatic, screen-detected persons (eContextual Question 2 in the Supplement).

This review has several limitations. First, it did not include the evidence regarding the diagnostic accuracy of screening tests for atrial fibrillation or the accuracy of a 12-lead ECG conducted and interpreted within primary care settings. A 2017 health technology assessment synthesized studies conducted in a variety of settings (eg, primary care, preoperative clinics, cardiology practices) that were related to diagnostic accuracy of ECG for atrial fibrillation.⁴⁹ Based on data from 7 studies, a 12-lead ECG interpreted by a nurse, general practitioner, or the ECG machine’s automated algorithm had a sensitivity of 92.7% (95% CI, 85.9%-96.8%) and specificity of 97.4% (95% CI, 95.0%-98.9%) when compared with a reference standard of cardiologist interpretation. The authors derived these estimates from a hierarchical summary receiver operating characteristics curve. Across individual studies, sensitivity ranged from 68% to 100% and specificity ranged from 76% to 100%.

Second, this review did not include the evidence on rate control or rhythm control for atrial fibrillation. Briefly, rhythm control is not recommended for asymptomatic adults with atrial fibrillation. Some guidelines, including those of the American Heart Association/American College of Cardiology/Heart Rhythm Society, recommend rate control to achieve a resting heart rate under 110 beats per minute for asymptomatic patients with atrial fibrillation.¹

Third, this review did not include head-to-head trials of treatments for atrial fibrillation because the intention was to provide evidence on benefits of treatments compared with placebo or no treatment rather than to assess the comparative effectiveness of treatments. Nevertheless, the review summarized a previously published network meta-analysis that provides comparative effectiveness estimates.

To better understand the potential benefits and harms of systematic screening for atrial fibrillation with ECG, randomized trials of asymptomatic persons that directly compare systematic screening with usual care and assess health outcomes are needed. The ongoing STROKESTOP study may help fill this evidence gap. Other relevant ongoing RCTs are focused on detecting atrial fibrillation (KQ2) as the primary outcome; these include SCREEN-AF,¹⁸ IDEAL-MD,⁵⁰ mSToPS,⁵¹ and D₂AF.⁵²

Although screening with ECG can detect previously unknown cases of atrial fibrillation, it has not been shown to detect more cases than screening focused on pulse palpation. Treatments for atrial fibrillation reduce the risk of stroke and all-cause mortality and increase the risk of bleeding, but trials have not assessed whether treatment of screen-detected asymptomatic older adults results in better health outcomes than treatment after detection by usual care or after symptoms develop.

Corresponding Author: Daniel E. Jonas, MD, MPH, University of North Carolina at Chapel Hill, 5034 Old Clinic Bldg, Chapel Hill, NC 27599 (daniel_jonas@med.unc.edu).

Accepted for Publication: March 19, 2018.

Author Contributions: Dr Jonas had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Jonas, Kahwati, Asher.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Jonas, Yun, Middleton, Coker-Schwimmer, Asher.

Critical revision of the manuscript for important intellectual content: Jonas, Kahwati, Yun, Coker-Schwimmer, Asher.

Statistical analysis: Jonas, Kahwati, Yun, Middleton, Asher.

Obtained funding: Jonas, Asher.

Administrative, technical, or material support: Kahwati, Middleton, Coker-Schwimmer.

Supervision: Jonas.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.

Funding/Support: This research was funded under contract HHSA-290-2015-00011-I, Task Order 7, from the Agency for Healthcare Research and Quality (AHRQ), US Department of Health and Human Services, under a contract to support the USPSTF.

Role of the Funder/Sponsor: Investigators worked with USPSTF members and AHRQ staff to develop the scope, analytic framework, and key questions for this review. AHRQ had no role in study selection, quality assessment, or synthesis. AHRQ staff provided project oversight, reviewed the report to ensure that the analysis met methodological standards, and distributed the draft for peer review. Otherwise, AHRQ had no role in the conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript findings. The opinions expressed in this document are those of the authors and do not reflect the official position of AHRQ or the US Department of Health and Human Services.

Additional Contributions: We gratefully acknowledge the following individuals for their contributions to this project, including AHRQ staff (Howard Tracer, MD, Elisabeth Kato, MD, and Tracy Wolff, MD) and RTI International–University of North Carolina Evidence-based Practice Center staff (Carol Woodell, BSPH; Josh Green, BA; Christiane Voisin, MSLS; Claire Korzen, BA; Sharon Barrell, MA; and Loraine Monroe). The USPSTF members, expert consultants, peer reviewers, and federal partner reviewers did not receive financial compensation for their contributions. Ms Woodell, Mr Green, Ms Voisin, Ms Korzen, Ms Barrell, and Ms Monroe received compensation for their role in this project.

Additional Information: A draft version of this evidence report underwent external peer review from 4 content experts (David A. Fitzmaurice, MD, University of Birmingham; Richard Hobbs, FRCP, Oxford University; Stephen Morgan, MBBChir [Cantab], University of Southampton; Larisa G. Tereshchenko, MD, Oregon Health & Science University) and 1 federal partner reviewer from the Centers for Disease Control and Prevention. Comments from reviewers were presented to the USPSTF during its deliberation of the evidence and were considered in preparing the final evidence review.

Editorial Disclaimer: This evidence report is presented as a document in support of the accompanying USPSTF Recommendation Statement. It did not undergo additional peer review after submission to JAMA.