Download as pdf or txt
Download as pdf or txt
You are on page 1of 3

Opinion Editorial

ARTICLE INFORMATION smokers: 5-year results of the MILD trial. Eur J 7. Kinsinger LS, Anderson C, Kim J, et al.
Author Affiliations: Department of Medicine, Cancer Prev. 2012;21(3):308-315. doi:10.1097/CEJ Implementation of lung cancer screening in the
Division of Cardiology, University of California, .0b013e328351e1b6 Veterans Health Administration. JAMA Intern Med.
San Francisco; Editor, JAMA Internal Medicine. 3. Infante M, Cavuto S, Lutman FR, et al; DANTE 2017;177(3):399-406. doi:10.1001/jamainternmed
Study Group. Long-term follow-up results of the .2016.9022
Corresponding Author: Rita F. Redberg, MD, MSc,
Department of Medicine, Division of Cardiology, DANTE trial, a randomized study of lung cancer 8. Caverly TJ, Fagerlin A, Wiener RS, et al.
University of California San Francisco, 505 screening with spiral computed tomography. Am J Comparison of observed harms and expected
Parnassus, M1180, San Francisco, CA 94143 Respir Crit Care Med. 2015;191(10):1166-1175. mortality benefit for persons in the Veterans Health
(rita.redberg@ucsf.edu). doi:10.1164/rccm.201408-1475OC Affairs Lung Cancer Screening Demonstration
4. Wille MMW, Dirksen A, Ashraf H, et al. Results of Project. JAMA Intern Med. 2018;178(3):426-428.
Published Online: August 13, 2018. doi:10.1001/jamainternmed.2017.8170
doi:10.1001/jamainternmed.2018.3527 the randomized Danish Lung Cancer Screening Trial
with focus on high-risk profiling. Am J Respir Crit 9. Goff SL, Mazor KM, Ting HH, Kleppel R,
Conflict of Interest Disclosures: Dr Redberg Care Med. 2016;193(5):542-551. doi:10.1164/rccm Rothberg MB. How cardiologists present the
chaired the April 2014 Medicare Evidence .201505-1040OC benefits of percutaneous coronary interventions to
Development & Coverage Advisory Committee patients with stable angina: a qualitative analysis.
meeting on lung cancer screening. 5. Huo J, Shen C, Volk RJ, Shih YT. Use of CT and
chest radiography for lung cancer screening before JAMA Intern Med. 2014;174(10):1614-1621.
and after publication of screening guidelines: doi:10.1001/jamainternmed.2014.3328
REFERENCES intended and unintended uptake. JAMA Intern Med. 10. Merchant FM, Dickert NW Jr, Howard DH.
1. Aberle DR, Adams AM, Berg CD, et al; National 2017;177(3):439-441. doi:10.1001/jamainternmed.2016 Mandatory shared decision making by the Centers
Lung Screening Trial Research Team. Reduced .9016 for Medicare & Medicaid Services for cardiovascular
lung-cancer mortality with low-dose computed 6. Brenner AT, Malo TL, Margolis M, et al. procedures and other tests [published online June
tomographic screening. N Engl J Med. 2011;365(5): Evaluating shared decision making for lung cancer 4, 2018]. JAMA. doi:10.1001/jama.2018.6617
395-409. doi:10.1056/NEJMoa1102873 screening [published online August 13, 2018]. JAMA
2. Pastorino U, Rossi M, Rosato V, et al. Annual or Intern Med. doi:10.1001/jamainternmed.2018.3054
biennial CT screening versus observation in heavy

Screening for Atrial Fibrillation Comes With Many Snags


John Mandrola, MD; Andrew Foy, MD; Gerald Naccarelli, MD

The association of atrial fibrillation (AF) with an increased risk


of stroke and heart failure makes it a serious health condi- Box 1. Sample Calculation of NNS for AF With ECG
tion. Many people have AF and do not know it, and its preva- to Prevent 1 Strokea
lence continues to rise in parallel with the growing numbers • New AF found on initial ECG in 0.5% of screened population
of people living with obesity and cardiac risk factors.1,2 • NNS to diagnose 1 new case of AF, 1/0.005 = 200 people
These elements, along with the proliferation of devices ca- • Estimated absolute risk reduction from anticoagulation, 2%
pable of monitoring the heart rhythm, have led many physi- • Number needed to treat to prevent 1 stroke, 1/0.02 = 50 people
cians to promote screening for AF.3 The hypothesis is that early • NNS to prevent 1 stroke, 200 × 50 = 10 000 people
detection would allow treatment with anticoagulation, which Abbreviations: AF, atrial fibrillation; ECG, electrocardiography; NNS, number
would then deliver a net clini- needed to screen.

cal benefit. Screening, how- a


All data are from the STROKESTOP study.7
Related articles at jama.com ever, is not a sure win. Two
pioneers of public health, Drs
Wilson and Jungner, wrote a are no trials that assess clinical outcomes from AF screening
half-century ago: “in theory…screening is an admirable method programs, we can only estimate efficacy—by multiplying the
of combating disease,…but in practice, there are snags.”4(p7) number needed to test to diagnose 1 new case of AF by the
A systematic review for the US Preventive Services Task number needed to treat with anticoagulation to prevent 1
Force (USPSTF) published in the current issue of JAMA,5 stroke.
conducted in support of the USPSTF Recommendation State- We used data from STROKESTOP,7 a study of AF screen-
ment published in the same issue,6 found the current evi- ing using intermittent ECG recordings among individuals aged
dence insufficient to assess the balance of benefits and harms 75 to 76 years in 2 Swedish regions. In this older-aged cohort,
of screening for AF with electrocardiography (ECG) (I state- the number needed to screen (NNS) to prevent 1 stroke equaled
ment). We concur with this conclusion and expand on many 10 000 (Box 1). For comparison, the NNS for abdominal aortic
of the “snags” of AF screening. aneurysm to save 1 disease-specific death is 250; for fecal oc-
cult blood testing screening, the NNS is 625.8 The prevalence
Lack of Efficacy of AF increases with age; thus, the NNS to prevent 1 stroke
An AF screening program must do more than pick up the ar- would be higher in people younger than 75 years.
rhythmia or increase use of anticoagulant drugs; it must de- A cost analysis is sobering. If an ECG costs $100, society
liver more benefit than harm in the screened group. Since there would pay $1 million (100 × 10 000 NNS) to prevent 1 stroke

1296 JAMA Internal Medicine October 2018 Volume 178, Number 10 (Reprinted) jamainternalmedicine.com

© 2018 American Medical Association. All rights reserved.


Downloaded from jamanetwork.com by KU Leuven 2Bergen Biomedical Library user on 04/21/2024
Editorial Opinion

and higher stroke risk, and stasis of blood in the atrium dur-
Box 2. Sample Calculation of AF Misdiagnosis at ECG Specificity ing AF provides biologic plausibility of a causal relationship.
of 95% But as Kamel et al14 have described, AF fails to fulfill many of
• ECG specificity, 95% the Hill criteria15 for judging causality.
• Screened population, 1 000 000 people Atrial fibrillation does not fit the temporality criteria of cau-
• AF prevalence, 2% sation. In the TRENDS study, the majority of patients who ex-
• Number with AF, 0.2 × 1 000 000 = 20 000 people perienced stroke had no AF episodes within the 30 days prior
• Number without AF, 1 000 000 − 20 000 = 980 000 people to their event.16 Atrial fibrillation also fails for specificity, in
• True negative, 0.95 × 980 000 = 931 000 people reassured
that many strokes occurring in patients with AF are caused by
they have no AF
• False positive, 0.05 × 980 000 = 49 000 people falsely either small- or large-vessel atherosclerosis, not cardiac
diagnosed with AF emboli.17,18 Atrial fibrillation may not comport with the bio-
logical gradient or dose-response curve; eg, there should be
Abbreviations: AF, atrial fibrillation; ECG, electrocardiography.
greater stroke risk from permanent AF than paroxysmal AF.
Some studies show a gradient of risk based on AF type19,20;
in persons older than 75 years, and even more in younger per- others do not.21,22 And if AF caused stroke, the Hill accor-
sons. Yet the financial burden of AF screening extends well be- dance with experimental evidence criteria15 would hold that
yond the mere cost of the ECG. The costs of medical care rhythm-control medicines that reduce AF should also reduce
associated with true- and false-positive diagnoses of AF as stroke risk. They do not.
well as incidental ECG findings, such as premature ventricu- Finally, and perhaps most importantly, is the false belief that
lar contractions, nonspecific ST-T wave changes, and brady- we know the risk of stroke in patients with untreated AF. A sys-
cardia, would surely be substantial. tematic review of 34 cohort studies and randomized clinical
trials enrolling patients not treated with anticoagulation found
Harms From AF Screening that stroke rates ranged from 0.45% to 9.28% per year.23 This
Tests for AF, whether by pulse palpation, ECGs, smartphone, variability was likely not due to baseline differences in co-
or watch recordings, are burdened by low specificity. Given the horts: the authors noted manifold variation of stroke rates across
low prevalence of AF in the populations screened, the ability any given CHA2DS2-VASc score. Provocatively, this review also
to correctly identify healthy patients without AF is critical. found that the Danish National Patient Registry, which is cited
In the Screening for AF in the Elderly (SAFE) study, the in AF treatment guidelines, had the second highest stroke rate,
specificity of ECG findings from general practitioners or nurses significantly higher than all but 1 of the other cohorts.
was approximately 90%.9 False-positive rates of AF detec-
tion in this range or greater have been confirmed with 12-lead Patients With Screen-Detected AF May Not Benefit From
ECGs, 10,11 the iPhone ECG, 12 and artificial intelligence– Anticoagulation as Symptomatic Patients in Older Trials Did
enhanced smart-watch detection of AF.13 There are no studies of patients treated for screening-detected
A test with a specificity of 90% may seem reasonable, but AF. In the 1990s, anticoagulation vs no therapy or aspirin for AF
its 10% rate of false positives can lead to large numbers of false was shown to provide a net clinical benefit in clinical trials that
diagnoses when used to search for a condition with low preva- enrolled patients mostly referred from physician offices.24 Thirty
lence. Even if screening had a generous 95% specificity, for ev- years later, however, the milieu has changed: stroke rates have
ery 1 million people screened, 49 000 people would be wrongly fallen,25,26 acute stroke therapy has improved; and the rise of
told they had AF (Box 2). If half of these people are started on AF detection devices has led to much more aggressive search
anticoagulation therapy, and the absolute risk increase of ma- for AF. Whether the net benefit of anticoagulation in patients
jor bleeding is 1%, then 250 people will experience unneces- with screen-detected AF remains as it was in the 1990s is a ques-
sary iatrogenic bleeding. tion that can only be answered with clinical trials.
In addition to bleeding, other harms from AF screening
include the anxiety of receiving a misdiagnosis of heart dis- Conclusions
ease, the risks of complications from the resulting additional If AF screening is adopted in the absence of any outcomes data,
diagnostic and therapeutic testing, and treatments related to hundreds of thousands, perhaps millions, of people will rightly
incidental findings on ECGs and misdiagnosed AF. and wrongly gain the diagnosis of a cardiac disorder. Before
turning this many people into patients, there should be com-
We Do Not Understand AF Well Enough to Screen for It pelling evidence that the benefits of this label exceed the
The second screening criterion of Wilson and Jungner calls for harms, and that these benefits can be achieved at an accept-
a solid understanding of the natural history of the condition.4 able cost. We propose and strongly encourage randomized
Atrial fibrillation is neither simple nor well understood. clinical trials of AF screening before adopting this practice.
We partially understand the relationship between AF and Experience tells us how hard it is to stop even a dangerous or
stroke: multiple cohort studies find an association between AF ineffective practice.

ARTICLE INFORMATION University College of Medicine, Hershey, Corresponding Author: John Mandrola, MD,
Author Affiliations: Baptist Health Louisville, Pennsylvania (Foy, Naccarelli). Baptist Health Louisville, 3900 Kresge Way,
Louisville, Kentucky (Mandrola); Penn State Louisville, KY 40207 (john.mandrola@gmail.com).

jamainternalmedicine.com (Reprinted) JAMA Internal Medicine October 2018 Volume 178, Number 10 1297

© 2018 American Medical Association. All rights reserved.


Downloaded from jamanetwork.com by KU Leuven 2Bergen Biomedical Library user on 04/21/2024
Opinion Editorial

Published Online: August 7, 2018. 9. Mant J, Fitzmaurice DA, Hobbs FDR, et al. in patients with various patterns of atrial fibrillation:
doi:10.1001/jamainternmed.2018.4038 Accuracy of diagnosing atrial fibrillation on results from the ENGAGE AF-TIMI 48 Trial (Effective
Conflict of Interest Disclosures: Dr Naccarelli has electrocardiogram by primary care practitioners Anticoagulation With Factor Xa Next Generation in
served as consultant/advisory board member for and interpretative diagnostic software: analysis of Atrial Fibrillation-Thrombolysis in Myocardial
Janssen, Daiichi Sankyo, GlaxoSmithKline, and Astra data from screening for atrial fibrillation in the Infarction 48). Circ Arrhythm Electrophysiol. 2017;10
Zeneca and has received research support from elderly (SAFE) trial. BMJ. 2007;335(7616):380. doi: (1):e004267. doi:10.1161/CIRCEP.116.004267
Janssen. No other disclosures are reported. 10.1136/bmj.39227.551713.AE 20. Ganesan AN, Chew DP, Hartshorne T, et al. The
10. Hwan Bae M, Hoon Lee J, Heon Yang D, et al. impact of atrial fibrillation type on the risk of
REFERENCES Erroneous computer electrocardiogram thromboembolism, mortality, and bleeding:
interpretation of atrial fibrillation and its clinical a systematic review and meta-analysis. Eur Heart J.
1. Lane DA, Skjøth F, Lip GYH, Larsen TB, Kotecha consequences. Clin Cardiol. 2012;35(6):348-353. 2016;37(20):1591-1602. doi:10.1093/eurheartj
D. Temporal trends in incidence, prevalence, and doi:10.1002/clc.22000 /ehw007
mortality of atrial fibrillation in primary care. J Am
Heart Assoc. 2017;6(5):e005155. doi:10.1161/JAHA 11. Bogun F, Anh D, Kalahasty G, et al. Misdiagnosis 21. Hohnloser SH, Pajitnev D, Pogue J, et al; ACTIVE
.116.005155 of atrial fibrillation and its clinical consequences. W Investigators. Incidence of stroke in paroxysmal
Am J Med. 2004;117(9):636-642. doi:10.1016/j versus sustained atrial fibrillation in patients taking
2. Foy AJ, Mandrola J, Liu G, Naccarelli GV. Relation .amjmed.2004.06.024 oral anticoagulation or combined antiplatelet
of obesity to new-onset atrial fibrillation and atrial therapy: an ACTIVE W Substudy. J Am Coll Cardiol.
flutter in adults. Am J Cardiol. 2018;121(9):1072-1075. 12. Lowres N, Neubeck L, Salkeld G, et al. Feasibility
and cost-effectiveness of stroke prevention 2007;50(22):2156-2161. doi:10.1016/j.jacc.2007.07
doi:10.1016/j.amjcard.2018.01.019 .076
through community screening for atrial fibrillation
3. Freedman B, Camm J, Calkins H, et al; AF-Screen using iPhone ECG in pharmacies: the SEARCH-AF 22. Flaker G, Ezekowitz M, Yusuf S, et al. Efficacy
Collaborators. Screening for atrial fibrillation: study. Thromb Haemost. 2014;111(6):1167-1176. and safety of dabigatran compared to warfarin in
a report of the AF-SCREEN International doi:10.1160/TH14-03-0231 patients with paroxysmal, persistent, and
Collaboration. Circulation. 2017;135(19):1851-1867. permanent atrial fibrillation: results from the RE-LY
doi:10.1161/CIRCULATIONAHA.116.026693 13. Tison GH, Sanchez JM, Ballinger B, et al. Passive
detection of atrial fibrillation using a commercially (Randomized Evaluation of Long-Term
4. Wilson J, Jungner G. Principles and Practice of available smartwatch. JAMA Cardiol. 2018;3(5): Anticoagulation Therapy) study. J Am Coll Cardiol.
Screening for Disease. Geneva, Switzerland: World 409-416. doi:10.1001/jamacardio.2018.0136 2012;59(9):854-855. doi:10.1016/j.jacc.2011.10.896
Health Organization; 1968. 23. Quinn GR, Severdija ON, Chang Y, Singer DE.
14. Kamel H, Okin PM, Elkind MSV, Iadecola C.
5. Jonas DE, Kahwati LC, Yun JDY, Middleton JC, Atrial fibrillation and mechanisms of stroke: time for Wide Variation in Reported Rates of Stroke Across
Coker-Schwimmer M, Asher GN. Screening for atrial a new model. Stroke. 2016;47(3):895-900. Cohorts of Patients With Atrial Fibrillation. Circulation.
fibrillation with electrocardiography: evidence 2017;135(3):208-219. doi:10.1161/CIRCULATIONAHA
report and systematic review for the US Preventive 15. Hill AB. The environment and disease: .116.024057
Services Task Force [published August 7, 2018]. JAMA. association or causation? Proc R Soc Med. 1965;58
(5):295-300. 24. Hart RG, Pearce LA, Aguilar MI. Meta-analysis:
doi:10.1001/jama.2018.4190 antithrombotic therapy to prevent stroke in
6. US Preventive Services Task Force. Screening for 16. Daoud EG, Glotzer TV, Wyse DG, et al; TRENDS patients who have nonvalvular atrial fibrillation.
atrial fibrillation with electrocardiography: US Investigators. Temporal relationship of atrial Ann Intern Med. 2007;146(12):857-867. doi:10.7326
Preventive Services Task Force Recommendation tachyarrhythmias, cerebrovascular events, and /0003-4819-146-12-200706190-00007
Statement [published August 7, 2018]. JAMA. systemic emboli based on stored device data:
a subgroup analysis of TRENDS. Heart Rhythm. 25. Kleindorfer DO, Khoury J, Moomaw CJ, et al.
doi:10.1001/jama.2018.10321 Stroke incidence is decreasing in whites but not in
2011;8(9):1416-1423. doi:10.1016/j.hrthm.2011.04.022
7. Svennberg E, Engdahl J, Al-Khalili F, Friberg L, blacks: a population-based estimate of temporal
Frykman V, Rosenqvist M. Mass screening for 17. Lodder J, Bamford JM, Sandercock PA, Jones trends in stroke incidence from the Greater
untreated atrial fibrillation: the STROKESTOP study. LN, Warlow CP. Are hypertension or cardiac Cincinnati/Northern Kentucky Stroke Study. Stroke.
Circulation. 2015;131(25):2176-2184. doi:10.1161 embolism likely causes of lacunar infarction? Stroke. 2010;41(7):1326-1331. doi:10.1161/STROKEAHA.109
/CIRCULATIONAHA.114.014343 1990;21(3):375-381. doi:10.1161/01.STR.21.3.375 .575043
8. Saquib N, Saquib J, Ioannidis JPA. Does 18. Chesebro JH, Fuster V, Halperin JL. Atrial 26. Aked J, Delavaran H, Norrving B, Lindgren A.
screening for disease save lives in asymptomatic fibrillation: risk marker for stroke. N Engl J Med. Temporal trends of stroke epidemiology in
adults? systematic review of meta-analyses and 1990;323(22):1556-1558. doi:10.1056 southern Sweden: a population-based study on
randomized trials. Int J Epidemiol. 2015;44(1): /NEJM199011293232209 stroke incidence and early case-fatality.
264-277. doi:10.1093/ije/dyu140 19. Link MS, Giugliano RP, Ruff CT, et al; ENGAGE Neuroepidemiology. 2018;50(3-4):174-182. doi:10
AF-TIMI 48 Investigators. Stroke and mortality risk .1159/000487948

1298 JAMA Internal Medicine October 2018 Volume 178, Number 10 (Reprinted) jamainternalmedicine.com

© 2018 American Medical Association. All rights reserved.


Downloaded from jamanetwork.com by KU Leuven 2Bergen Biomedical Library user on 04/21/2024

You might also like