CHA2DS2VASc Score and Adverse Outcomes in Middle-Aged Individuals Without Atrial

Fibrillation

Giulia Renda1, Fabrizio Ricci2,3, Giuseppe Patti4, Nay Aung5,6, Steffen E Petersen5,6, Sabina

Gallina1,2, Viktor Hamrefors3,7, Olle Melander3,7, Richard Sutton8, Gunnar Engstrom3, Raffaele De

Caterina1, Artur Fedorowski3,9

1 Institute of Cardiology, Department of Neuroscience, Imaging and Clinical Sciences, and Center

of Excellence on Aging, CeSI-Met, G. d’Annunzio University, Chieti, Italy

2 Institute of Advanced Biomedical Technologies, Department of Neuroscience, Imaging and

Clinical Sciences, “G. d’Annunzio” University, Chieti, Italy

3 Department of Clinical Sciences, Malmö, Faculty of Medicine, Lund University, Clinical

Research Center, 214 28 Malmö, Sweden.

4 Chair of Cardiology, University of L’Aquila, Italy.

5William Harvey Research Institute, NIHR Barts Biomedical Research Centre, Queen Mary

University of London, Charterhouse Square, London, UK.

6 Barts Heart Centre, St Bartholomew’s Hospital, Barts Health NHS Trust, West Smithfield,

London, UK.

7 Department of Internal Medicine, Skåne University Hospital, 214 28 Malmö, Sweden.

8 National Heart and Lung Institute, Imperial College, Hammersmith Hospital Campus, Ducane

Road, London W12 0NN, UK.

9 Department of Cardiology, Skåne University Hospital, 214 28 Malmö, Sweden.

1

Corresponding Author:

Giulia Renda, MD, PhD

Department of Neurosciences, Imaging and Clinical Sciences

Experimental Cardiology – CeSI-MeT

G. d’Annunzio University, Chieti-Pescara

Via L. Polacchi 11, 66100 Chieti – tel./fax 0871-541327

Institute of Cardiology - SS. Annunziata Hospital, Chieti

Via dei Vestini 31, 66100 Chieti - tel. 0871-41512, fax: 0871-402817

E-mail: grenda@unich.it

Word count (including references and figure legends): 4226

2

ABSTRACT

Aims: To assess the prognostic yield of CHA2DS2VASc score for new-onset atrial fibrillation (AF),

cardiovascular (CV) morbidity and mortality in a non-AF population.

Methods and Results: We analysed a population-based cohort of 22,179 middle-aged individuals

(mean age, 637 years; men, 39%). We grouped the population into five CHA2DS2VASc score

strata (0-1-2-3-≥4), and compared the risk of major adverse cerebro-cardiovascular events

(MACCE) and mortality between subjects with (n=3,542) and without history of AF (n=18,637).

Further, we analysed the annual incidence of AF across different CHA2DS2VASc strata.

Over a median (interquartile range) follow-up of 15 (2.9) years, 1,572 patients (6.9%) had

ischaemic strokes, 2,162 (9.5%) coronary events, and 5,899 (26%) died. The cumulative incidence

of ischaemic stroke in CHA2DS2VASc≥4 subjects without AF was similar to patients with AF and

CHA2DS2VASc=2, with a 10-year crude incidence rate of 0.91 (95%CI 0.68-1.19) and 1.13 (95%CI

0.93-1.36) ischaemic strokes per 100 patient-year, respectively. CHA2DS2VASc in a non-AF

population showed higher predictive accuracy for ischaemic stroke compared with AF population

(AUC=60.4% vs AUC=55.7%; p=0.001). In multivariable Cox regression analysis,

CHA2DS2VASc≥2 was an independent predictor of all-cause mortality (adjusted Hazard Ratio

(aHR): 1.58; 95%CI 1.48-1.69), CV death (aHR: 1.77; 95%CI 1.57-1.99), ischaemic stroke (aHR:

2.01; 95%CI 1.84-2.36) and coronary events (aHR 1.68; 95%CI 1.51-1.87). The cumulative

incidence of AF was greater with increasing CHA2DS2VASc strata, with an absolute annual

incidence >2%/year if CHA2DS2VASc≥4.

Conclusion: The CHA2DS2VASc score is a sensitive tool for predicting new-onset AF and adverse

outcomes in subjects both with and without atrial fibrillation.

Abstract Word Count: 2493

Key words: atrial fibrillation; ischaemic stroke; thromboembolic risk; anticoagulation;

cardiovascular risk.

Funding: This work was supported by grants from the Swedish Medical Research Council, the

Swedish Heart and Lung Foundation, the Medical Faculty of Lund University, Malmö University

Hospital, the Albert Påhlsson Research Foundation, the Crafoord Foundation, the Ernhold

Lundströms Research Foundation, the Region Skane, the Hulda and Conrad Mossfelt Foundation,

the King Gustaf V and Queen Victoria Foundation, The Wallenberg Foundation and the Lennart

Hanssons Memorial Fund. SEP and NA acknowledge the support through the Barts Biomedical

Research Centre which is supported and funded by the National Institute for Health Research.

4

BACKGROUND

The CHA2DS2-VASc score (congestive heart failure, hypertension, age ≥75 years [doubled],

diabetes, stroke/transient ischaemic attack/thromboembolism [doubled], vascular disease [prior

myocardial infarction, peripheral artery disease, or aortic plaque], age 65-74 years, sex category

[female]), is used to evaluate the risk of thromboembolic events in patients with non-valvular atrial

fibrillation (AF), and is recommended by current guidelines to guide the management of

antithrombotic therapy1. Most of the individual components of the CHA2DS2-VASc score have been

considered as risk factors for stroke, irrespective of the presence of AF, and are also included in

older stroke risk assessment models, such as the Framingham risk score 2; furthermore, they have

been considered as cardiovascular risk factors in other risk assessment tools that focus on the

prediction of overall cardiovascular risk 3. Many components of CHA2DS2-VASc score have also

been shown to predict outcome after stroke. In particular, heart failure, hypertension, age, diabetes,

and previous stroke have been related with worse outcome in patients with acute stroke, irrespective

of the presence of AF 4-6.

In recent years, the CHA2DS2-VASc score has been therefore proposed to predict cardiovascular

outcomes in clinical settings different from known AF. The score has been related to subsequent all-

cause mortality 7, recurrent stroke, cardiovascular events, and total mortality 8 in stroke patients;

with subsequent stroke and death after an acute coronary event9 or after coronary artery bypass

grafting 10; with subsequent ischaemic stroke, thromboembolism, and death in heart failure 11; and

with subsequent stroke in a population free from AF burden 12. Since the CHA2DS2-VASc score is

user-friendly and easy to apply, it might be useful for prospective risk assessment of major adverse

cardio- and cerebro-vascular events (MACCE) in the non-AF population, possibly guiding

therapeutic decisions. Moreover, as many of the individual risk factors included in this score are

also risk factors for AF itself, it might serve as a screening tool for the identification of patients at

higher risk of new-onset AF. 5

Accordingly, we hypothesised that CHA2DS2-VASc score may be independently associated with

MACCE and mortality in an AF-free population; and patients without AF but with high CHA2DS2-

VASc score may have risk of ischaemic stroke reaching/crossing the threshold for anticoagulation.

Finally, we sought to assess the performance of the CHA2DS2-VASc score in predicting new-onset

AF in subjects without a pre-existing diagnosis of AF.

METHODS

Study cohort

The Malmö Diet and Cancer Study is a prospective cohort study in which all men born between

1923 and 1945 and women born between 1923 and 1950 from the city of Malmö, Sweden (total

population: 330,000) were invited to participate. The participation rate was ~40%. Men and women,

totalling 30,446 participants, underwent a baseline examination between 1991 and 1996. Of these,

22,369 underwent a five-year rescreening based on a follow-up questionnaire in 1997-2001. The

average follow-up time from the rescreening was 15±4 years (Figure 1). Full description of

recruitment and screening procedures have been provided elsewhere13. The study complied with the

Declaration of Helsinki, and the protocol was approved by the regional ethics committee.

Definition of clinical characteristics

The participants underwent measurement of body weight and height, blood pressure (BP), and filled

a questionnaire on health, lifestyle, socio-economic factors and medications at the baseline

examination. Body surface area was calculated using the DuBois formula. Body-mass index (BMI)

was calculated as weight in kilograms divided by the square of height in meters (kg/m2). Baseline

questionnaire recorded the smoking status. Hypertension was defined as systolic BP≥140 mmHg

and/or diastolic BP≥90 mmHg or use of antihypertensive medications. Diabetes was defined as self-6

reported, physician diagnosis of diabetes or use of anti-diabetic medications. History of AF, heart

failure, stroke and vascular disease (previous myocardial infarction, any myocardial

revascularisation, and/or peripheral artery disease) were defined as self-reported, physician

diagnosis or based upon case-retrieval from the Swedish National Hospital Discharge Register

(SNHDR).

CHA2DS2-VASc score has been calculated based on the presence of its components retrieved both

at baseline and at five-year rescreening in order to validate the stability of the score itself.

Ascertainment of clinical endpoints

All study participants were followed-up through December 31, 2014 by linking a unique 10-digit

personal identification number with SNHDR, Swedish National Cause of Death Register (SNCDR)

and Stroke Register of Malmö (STROMA). Diagnoses in the SNHDR are coded as primary or

contributory and in the SNCDR as underlying or contributory cause of death, both using the

International Classification of Disease (ICD). The 9th edition (ICD-9) was used between 1987 and

1996 and the 10th edition (ICD-10) from 1997 until present. Event-free subjects (n=116; 0.5%) who

emigrated from Sweden before 31 December 2014 were assigned date of emigration as the last

follow-up date. A first diagnosis of AF or atrial flutter was ascertained by linkage of Swedish

personal identification numbers to SNHDR and SDCDR using diagnosis codes 427D for ICD-9 and

I48 for ICD-10. As in previous studies14, 15, the outcome of AF was defined as either a diagnosis of

AF or atrial flutter, given the close interrelationship of these diseases. Subjects with a hospital

discharge diagnosis of HF (427.00, 427.10, 428 [ICD-9] and I50 [ICD-10]) were considered to have

HF if the diagnosis was listed as the primary diagnosis.

Furthermore, we analysed the following outcomes: all-cause mortality, CV death, any stroke,

ischaemic stroke and coronary event. Stroke was defined according to the International

Classification of Diseases (ICD) 9 and 10 systems as cases coded 430, 431, 434, and 436 or I60,

I61, I63, and I64, respectively. STROMA was used for case retrieval. In addition, SNHDR and 7

SNCDR were used for retrieval of patients who moved out of Malmö. Ischaemic stroke was

ascertained by ICD9-434 and ICD10-I63 codes. Coronary event was defined as fatal or non-fatal

myocardial infarction or death due to coronary heart disease according to ICD9-410 and ICD10-I21

codes in SNHDR, and codes 410, 412, and 414 (ICD9) or I21–I23 and I25 (ICD10) in SNCDR. The

register-based diagnosis of stroke and coronary event in STROMA and SNHDR have been found to

be highly valid16.

Statistical analysis

Baseline characteristics of the study population at the time of 1997-2001 rescreening were

described using means (standard deviations [SD]) for continuous measures and number

(percentage) for categorical measures. For the purpose of our analysis, we defined the AF

population as subjects with prevalent and/or incident AF cases, and the non-AF population as

subjects who never experienced AF from baseline until the end of the follow-up. We used unpaired

Student’s t test to compare AF (prevalent or incident) versus non-AF groups for continuous

variables and Fisher’s exact test for categorical variables. We calculated crude incidence rates of all

end-points stratified by the presence of AF. Ninety-five percent confidence intervals (95% CIs) of

the crude incidence rates were estimated with exact method of Poisson distribution. We performed

time-to-event analyses by Kaplan-Meier estimator and Bonferroni-corrected log-rank test. We

computed the absolute risks of all end-points while accounting for competing risk of death using the

method recommended by Gray17. Relative risks according to CHA2DS2-VASc score were estimated

using the Jackknife pseudo-values modelled in generalised estimating equation to account for

competing risk of death. Hazard ratios with 95% CIs were derived from Cox models and the

proportional hazards assumption was verified with Shoenfeld residuals. Violation of proportional

hazards assumptions by time-varying covariates was handled by introducing interaction-terms

containing the starting time18. The predictive accuracy of CHA2DS2-VASc score from logistic

regression modelling was determined with receiver operating characteristic (ROC) analysis and the 8

resultant areas-under-the-curve (AUCs) in AF and non-AF groups were compared using the Hanley

and McNeil method19. A p-value of < 0.05 was considered significant. Survival analyses were

performed using ‘survival’, ‘survminer’, and ‘cmprsk’, ‘prodlim’, and ‘pROC’ packages and all

statistical analyses were done in R (R Core Team 2018; R: A language and environment for

statistical computing. R Foundation for Statistical Computing, Vienna, Austria; URL:

https://www.R-project.org).

RESULTS

The selection of study population is shown in Figure 1. The population-based prospective cohort

(Table 1) included 22,179 middle-aged individuals (mean age 637 years; men 39%). The median

(IQR) follow-up period was 15 (2.9) years. Baseline characteristics of the subjects, divided in 2

groups according to the presence or absence (prevalent or incident) of AF are reported in Table 1.

Mean CHA2DS2-VASc score was 1.4 in subjects without AF and 1.9 in those with AF. The

distribution of CHA2DS2-VASc score in the 2 groups is shown in Table 1.

CHA2DS2-VASc score and risk stratification for MACCE and mortality in subjects with and

without AF

During the follow-up period, 4406 (20%) subjects had a MACCE, 1899 (8.6%) had a stroke,

including 1572 (7.1%) with an ischaemic stroke, 2162 (9.7%) a coronary event, and 5899 (26%)

died (8.5% CV death). The crude incidence rates of considered events at the end of follow-up,

according to the CHA2DS2-VASc strata, in the 2 groups of subjects are reported in Table 2.

Cumulative incidences, absolute and relative risks of MACCE and mortality were greater with

increasing CHA2DS2VASc strata in subjects with and without AF (see OS Figure 1 for cumulative

incidences of all stroke and coronary events, and OS Figure 2 for CV and all-cause mortality). In

particular, the cumulative incidence of ischaemic stroke in the 2 groups of subjects, according to 9

CHA2DS2VASc strata, is shown in Figure 2, panel A (subjects without AF) and panel B (subjects

with AF). Subjects with CHA2DS2VASc≥4 without AF had a similar incidence of ischaemic stroke

to that of patients with AF and CHA2DS2VASc=2 (Figure 3, panel A), particularly during the first

10 years of follow-up, with a crude incidence rate of 0.91 (95% CI 0.68-1.19) and 1.13 (95% CI

0.93-1.36) ischaemic strokes per 100 patient-year, respectively (Figure 3, panel B).

Absolute risks of ischaemic stroke at different time points across different CHA2DS2VASc strata in

both subjects with and without AF are displayed in OS Figure 3. Notably, CHA2DS2VASc≥4 in

patients without AF carried a 5-year absolute risk of ischaemic stroke similar to CHA2DS2VASc=2

in patients with AF (4.4% vs 4.3%, p= NS).

CHA2DS2VASc in the non-AF population showed a higher predictive accuracy for ischaemic stroke

compared with AF population (AUC= 60.4% vs AUC= 55.7%; p= 0.001).

In the multivariate Cox regression analysis, a CHA2DS2VASc≥2 was a significant predictor of all-

cause death (HR 1.58; 95%CI 1.48-1.69), CV death (HR 1.77; 95%CI 1.57-1.99), ischaemic stroke

(HR 2.01; 95%CI 1.84-2.36), and coronary events (HR 1.68; 95%CI 1.51-1.87), regardless of AF,

age at the event, sex, use of antithrombotic drugs, dyslipidaemia and smoking status (Table 3).

CHA2DS2-VASc score and risk stratification for incident AF

Overall, 3094 (13.9%) new AF cases were detected, accounting for an absolute annual incidence of

1.0% (0.99-1.07)/year. The cumulative incidence of AF was greater with increasing CHA2DS2VASc

strata (Figure 4), with an absolute annual incidence of 2.5%/year among individuals with

CHA2DS2VASc≥4.

CHA2DS2VASc had acceptable discrimination performance (C-statistic=0.61) for predicting new-

onset AF.

Multivariable Cox regression analysis revealed CHA2DS2VASc≥2 as an independent predictor of

incident AF (HR 1.61; 95%CI 1.47-1.76), regardless of age at event, hyperlipidaemia, smoking

status and antithrombotic therapy.10

DISCUSSION

In this study, we have shown that:

1) The CHA2DS2VASc score enables mortality and cardiovascular risk stratification in non-AF

middle-aged adults;

2) A CHA2DS2VASc≥4 in non-AF subjects is associated with a 5-year absolute risk of

ischaemic stroke similar to CHA2DS2VASc=2 in AF subjects;

3) A CHA2DS2VASc≥4 heralds an ischaemic stroke rate >0.9%/year among non-AF subjects,

matching a previously proposed threshold for the safe prescription of non-vitamin K oral

anticoagulants in patients with AF;

4) The CHA2DS2VASc score predicts the risk of incident AF in a middle-aged population.

The CHA2DS2VASc score allows a simple and rapid assessment of thromboembolic risk in patients

with non-valvular AF according to the presence of a number of risk factors such as congestive heart

failure, hypertension, age, diabetes, stroke/transient ischaemic attack/thromboembolism, vascular

disease, and sex category. Most of the individual components of the CHA2DS2-VASc score are

established risk factors for stroke, irrespective of the presence of AF. Accordingly, longitudinal

prospective studies showed that CHA2DS2-VASc score is a predictor of ischaemic stroke and death

in high-risk patients without known AF, such as patients with heart failure11 or coronary heart

disease10. Our study confirms and further extends this evidence from a population-based prospective

cohort of 22,179 middle-aged individuals, where CHA2DS2VASc predicted the risk of ischaemic

stroke, any stroke, coronary event, cardiovascular and all-cause mortality, regardless of the AF

status, during a particularly long follow-up.

We observed that in AF-free subjects with a CHA2DS2VASc≥4 the risk of ischaemic stroke was

similar to AF subjects with CHA2DS2VASc=2. However, this became less evident after longer (>10 11

years) follow-up; this is likely driven by age difference at baseline, as AF individuals were

significantly older than non-AF population, and the potential for multiple unmeasured confounders.

Patients with AF and CHA2DS2VASc score ≥2 are considered to benefit most from anticoagulant

therapy. Notably, CHA2DS2VASc≥4 in non-AF subjects indicates an ischaemic stroke rate

>0.9%/year, which exactly matches the previously proposed “tipping point” 20 above which the

prescription of NOAC therapy is considered to be effective and safe in patients with AF.

Recent data from the COMPASS trial showed that patients with stable atherosclerotic vascular

disease and without known AF treated with a combination of low-dose aspirin (100 mg/day) and

low-dose rivaroxaban (2.5 mg twice daily) had a lower risk of major acute vascular events,

including ischaemic stroke, and a lower risk of death, when compared with low-dose aspirin alone21.

This benefit appears to be particularly striking for subsequent stroke (allegedly atherothrombotic

stroke in the majority of such patients without AF). This evidence stands in line with the recent

results from the COMMANDER HF trial suggesting a potential additional benefit of rivaroxaban

2.5 mg twice daily in the reduction of stroke among HF patients without prevalent AF22. From these

data, we may infer that oral anticoagulation exerts a significant protective effect against ischaemic

cerebrovascular events, regardless of the AF status. Further, a recent analysis suggested that the

availability of non-vitamin K antagonist oral anticoagulants (NOAC), with a better efficacy and

safety profile compared with vitamin-K antagonists (VKA) lowered the threshold for

anticoagulation to a stroke rate level of 0.9%/year in patients with history of AF 20.

Our findings provide a reasonable basis for further research aimed to test the efficacy of NOAC in

the primary prevention of ischaemic cerebrovascular events, regardless of AF status and the

underlying pathological mechanisms of stroke, namely cardioembolic or atherothrombotic. This is,

also, in line with previous evidence from the WARIS II trial, where oral anticoagulation, with

warfarin, was associated with significant reduction of ischaemic stroke in patients with a recent

acute myocardial infarction, regardless of the AF status 23. Conversely, the Phase III NAVIGATE-

ESUS study has recently shown no benefit with rivaroxaban at a daily dose of 15 mg, compared 12

with aspirin in a daily dose of 100 mg, for the prevention of stroke recurrence in patients with

embolic stroke of undetermined source24. Further investigations - based on post-hoc analyses of

aforementioned randomized controlled trials - should probably look at efficacy and safety outcomes

of oral anticoagulation in patients without known AF across different CHA2DS2-VASc score strata.

Detection of asymptomatic AF provide an opportunity to prevent ischaemic stroke by instituting

appropriate anticoagulation25. Opportunistic screening in all patients contacting the health system

≥65 years of age has been adopted in the ESC AF guidelines1, but might be more efficient if an

older age threshold is chosen or an additional stroke risk factor selected26. Despite the bulk of

investigations aimed to conceive the most appropriate tool and setting for AF screening, also

including use of cardiac implanted electronic devices (CIED) to detect atrial high-rate episodes

(AHRE), little is known about which individuals would benefit most from a widespread AF

screening programme, and which screening technique and risk stratification tool would provide the

optimal diagnostic yield. The CHA2DS2-VASc score, a simple, practical and user-friendly scoring

scheme which was initially used for stroke risk discrimination among patients with AF has gained

attention for predicting different cardiovascular outcomes in different populations. Importantly, a

retrospective analysis from a large nationwide Taiwanese dataset of nearly 70 000 patients with

type 2 diabetes mellitus, reported on the predictive ability of CHA2DS2-VASc score for stratifying

risk of incident AF27.

Compared with the strategies for “retrospective” identification of AF and thromboembolic risk

prevention, the “prospective” risk analysis for new-onset AF has gained hitherto relatively little

attention. Our data strengthen current evidence about the concurrent ability of CHA2DS2-VASc

score for predicting new-onset AF and stratifying the risk of stroke, coronary events, CV mortality,

and all-cause mortality in non-AF individuals. This may prompt the idea of rethinking current

strategies of cardiovascular risk prediction and consider AF itself as an additive risk factor28 - on top

of other CHA2DS2-VASc components - for stroke and major cardiovascular events. 13

Strengths and limitations

The principal strengths of this study are the large study population, the extensive follow-up time, as

well as the reliability of the prospective data collection protocol of our registries.

The main limitation of this study is the lack of prospective reappraisal of CHA2DS2-VASc risk

factors and assessment of oral antithrombotic therapy during follow-up, as well as the lack of

important pieces of information such as AF classification and bleeding events.

Conclusions

The CHA2DS2VASc score is a sensitive tool for prediction of new-onset AF and risk stratification

of MACCE in subjects both with and without prevalent or incident AF. The annual rate of

ischaemic stroke in CHA2DS2VASc≥4 patients without AF was above 0.9%/year, exceeding current

advised threshold for safe prescription of non-vitamin K oral anticoagulants in patients with atrial

fibrillation. Pending further validation studies, the presence of AF may be considered as equivalent

to two additional points in the CHA2DS2VASc score for stroke prediction.

14

REFERENCES

1. Kirchhof P, Benussi S, Kotecha D, Ahlsson A, Atar D, Casadei B, Castella M, Diener HC, Heidbuchel H, Hendriks J, Hindricks G, Manolis AS, Oldgren J, Popescu BA, Schotten U, Van Putte B, Vardas P, Group ESCSD. 2016 ESC Guidelines for the management of atrial fibrillation developed in collaboration with EACTS. Eur Heart J 2016;37(38):2893-2962.2. Wolf PA, D'Agostino RB, Belanger AJ, Kannel WB. Probability of stroke: a risk profile from the Framingham Study. Stroke 1991;22(3):312-8.3. Collins GS, Altman DG. An independent external validation and evaluation of QRISK cardiovascular risk prediction: a prospective open cohort study. BMJ 2009;339:b2584.4. Ntaios G, Faouzi M, Ferrari J, Lang W, Vemmos K, Michel P. An integer-based score to predict functional outcome in acute ischemic stroke: the ASTRAL score. Neurology 2012;78(24):1916-22.5. Vemmos K, Ntaios G, Savvari P, Vemmou AM, Koroboki E, Manios E, Kounali A, Lip GY. Stroke aetiology and predictors of outcome in patients with heart failure and acute stroke: a 10-year follow-up study. Eur J Heart Fail 2012;14(2):211-8.6. Weimar C. Stroke: initial stroke volume is an independent outcome predictor. Nat Rev Neurol 2012;8(6):305-6.7. Henriksson KM, Farahmand B, Johansson S, Asberg S, Terent A, Edvardsson N. Survival after stroke--the impact of CHADS2 score and atrial fibrillation. Int J Cardiol 2010;141(1):18-23.8. Ntaios G, Lip GY, Makaritsis K, Papavasileiou V, Vemmou A, Koroboki E, Savvari P, Manios E, Milionis H, Vemmos K. CHADS(2), CHA(2)S(2)DS(2)-VASc, and long-term stroke outcome in patients without atrial fibrillation. Neurology 2013;80(11):1009-17.9. Mitchell LB, Southern DA, Galbraith D, Ghali WA, Knudtson M, Wilton SB, investigators A. Prediction of stroke or TIA in patients without atrial fibrillation using CHADS2 and CHA2DS2-VASc scores. Heart 2014;100(19):1524-30.10. Biancari F, Asim Mahar MA, Kangasniemi OP. CHADS2 and CHA2DS2-VASc scores for prediction of immediate and late stroke after coronary artery bypass graft surgery. J Stroke Cerebrovasc Dis 2013;22(8):1304-11.11. Melgaard L, Gorst-Rasmussen A, Lane DA, Rasmussen LH, Larsen TB, Lip GY. Assessment of the CHA2DS2-VASc Score in Predicting Ischemic Stroke, Thromboembolism, and Death in Patients With Heart Failure With and Without Atrial Fibrillation. JAMA 2015;314(10):1030-8.12. Lip GY, Lin HJ, Chien KL, Hsu HC, Su TC, Chen MF, Lee YT. Comparative assessment of published atrial fibrillation stroke risk stratification schemes for predicting stroke, in a non-atrial fibrillation population: the Chin-Shan Community Cohort Study. Int J Cardiol 2013;168(1):414-9.13. Smith JG, Platonov PG, Hedblad B, Engstrom G, Melander O. Atrial fibrillation in the Malmo Diet and Cancer study: a study of occurrence, risk factors and diagnostic validity. Eur J Epidemiol 2010;25(2):95-102.14. Benjamin EJ, Levy D, Vaziri SM, D'Agostino RB, Belanger AJ, Wolf PA. Independent risk factors for atrial fibrillation in a population-based cohort. The Framingham Heart Study. JAMA 1994;271(11):840-4.15. Psaty BM, Manolio TA, Kuller LH, Kronmal RA, Cushman M, Fried LP, White R, Furberg CD, Rautaharju PM. Incidence of and risk factors for atrial fibrillation in older adults. Circulation 1997;96(7):2455-61.16. Ludvigsson JF, Andersson E, Ekbom A, Feychting M, Kim JL, Reuterwall C, Heurgren M, Olausson PO. External review and validation of the Swedish national inpatient register. BMC Public Health 2011;11:450.17. Gray RJ. A Class of K-Sample Tests for Comparing the Cumulative Incidence of a Competing Risk. The Annals of Statistics 1988;16(3):1141-1154.18. Bellera CA, MacGrogan G, Debled M, de Lara CT, Brouste V, Mathoulin-Pelissier S. Variables with time-varying effects and the Cox model: some statistical concepts illustrated with a prognostic factor study in breast cancer. BMC Med Res Methodol 2010;10:20.19. Hanley JA, McNeil BJ. A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology 1983;148(3):839-43.

15

20. Eckman MH, Singer DE, Rosand J, Greenberg SM. Moving the tipping point: the decision to anticoagulate patients with atrial fibrillation. Circ Cardiovasc Qual Outcomes 2011;4(1):14-21.21. Eikelboom JW, Connolly SJ, Bosch J, Dagenais GR, Hart RG, Shestakovska O, Diaz R, Alings M, Lonn EM, Anand SS, Widimsky P, Hori M, Avezum A, Piegas LS, Branch KRH, Probstfield J, Bhatt DL, Zhu J, Liang Y, Maggioni AP, Lopez-Jaramillo P, O'Donnell M, Kakkar AK, Fox KAA, Parkhomenko AN, Ertl G, Stork S, Keltai M, Ryden L, Pogosova N, Dans AL, Lanas F, Commerford PJ, Torp-Pedersen C, Guzik TJ, Verhamme PB, Vinereanu D, Kim JH, Tonkin AM, Lewis BS, Felix C, Yusoff K, Steg PG, Metsarinne KP, Cook Bruns N, Misselwitz F, Chen E, Leong D, Yusuf S, Investigators C. Rivaroxaban with or without Aspirin in Stable Cardiovascular Disease. N Engl J Med 2017;377(14):1319-1330.22. Zannad F, Anker SD, Byra WM, Cleland JGF, Fu M, Gheorghiade M, Lam CSP, Mehra MR, Neaton JD, Nessel CC, Spiro TE, van Veldhuisen DJ, Greenberg B, Investigators CH. Rivaroxaban in Patients with Heart Failure, Sinus Rhythm, and Coronary Disease. N Engl J Med 2018.23. Hurlen M, Abdelnoor M, Smith P, Erikssen J, Arnesen H. Warfarin, aspirin, or both after myocardial infarction. N Engl J Med 2002;347(13):969-74.24. Hart RG, Sharma M, Mundl H, Kasner SE, Bangdiwala SI, Berkowitz SD, Swaminathan B, Lavados P, Wang Y, Wang Y, Davalos A, Shamalov N, Mikulik R, Cunha L, Lindgren A, Arauz A, Lang W, Czlonkowska A, Eckstein J, Gagliardi RJ, Amarenco P, Ameriso SF, Tatlisumak T, Veltkamp R, Hankey GJ, Toni D, Bereczki D, Uchiyama S, Ntaios G, Yoon BW, Brouns R, Endres M, Muir KW, Bornstein N, Ozturk S, O'Donnell MJ, De Vries Basson MM, Pare G, Pater C, Kirsch B, Sheridan P, Peters G, Weitz JI, Peacock WF, Shoamanesh A, Benavente OR, Joyner C, Themeles E, Connolly SJ, Investigators NE. Rivaroxaban for Stroke Prevention after Embolic Stroke of Undetermined Source. N Engl J Med 2018.25. Healey JS, Connolly SJ, Gold MR, Israel CW, Van Gelder IC, Capucci A, Lau CP, Fain E, Yang S, Bailleul C, Morillo CA, Carlson M, Themeles E, Kaufman ES, Hohnloser SH, Investigators A. Subclinical atrial fibrillation and the risk of stroke. N Engl J Med 2012;366(2):120-9.26. Benito L, Coll-Vinent B, Gomez E, Marti D, Mitjavila J, Torres F, Miro O, Siso A, Mont L. EARLY: a pilot study on early diagnosis of atrial fibrillation in a primary healthcare centre. Europace 2015;17(11):1688-93.27. Hu WS, Lin CL. Role of CHA2DS2-VASc score in predicting new-onset atrial fibrillation in patients with type 2 diabetes mellitus with and without hyperosmolar hyperglycaemic state: real-world data from a nationwide cohort. BMJ Open 2018;8(3):e020065.28. Brambatti M, Connolly SJ, Gold MR, Morillo CA, Capucci A, Muto C, Lau CP, Van Gelder IC, Hohnloser SH, Carlson M, Fain E, Nakamya J, Mairesse GH, Halytska M, Deng WQ, Israel CW, Healey JS, Investigators A. Temporal relationship between subclinical atrial fibrillation and embolic events. Circulation 2014;129(21):2094-9.

16

TABLES

NO AF (n=18,367)

AF (n=3,542)

P

Age, mean (SD) 62 (7.4) 67 (6.7) <0.001Male sex, n (%) 6770 (36.3) 1836 (51.8) <0.001BMI, mean (SD) 25 (3.8) 27 (4.1) <0.001Hypertension, n (%) 3568 (19.1) 1057 (29.8) <0.001Diabetes, n (%) 692 ( 3.7) 216 ( 6.1) <0.001Heart Failure, n (%) 282 ( 1.5) 312 ( 8.8) <0.001Smoking history, n (%) 5080 (27.3) 791 (22.3) <0.001Prevalent CVD, n (%) 471 ( 2.5) 431 (12.2) <0.001Previous Stroke, n (%) 318 ( 1.7) 153 ( 4.3) <0.001Previous cancer, n (%) 1068 ( 5.7) 246 ( 6.9) 0.006Previous PCI, n (%) 18 ( 0.1) 6 ( 0.2) 0.353Previous CABG, n (%) 212 ( 1.1) 23 ( 0.6) 0.012Drug Therapy ACE-inhibitors, n (%) 512 ( 2.7) 108 ( 3.0) 0.345 Beta-blockers, n (%) 1835 ( 9.8) 357 (10.1) 0.693 Diuretics, n (%) 1072 ( 5.8) 210 ( 5.9) 0.708 Calcium hannel blockers, n (%) 913 ( 4.9) 165 ( 4.7) 0.570 Hypolipidemics, n (%) 575 ( 3.1) 94 ( 2.7) 0.186 Antidiabetics, n (%) 279 ( 1.5) 51 ( 1.4) 0.856 Antithrombotics, n (%) 1547 ( 8.3) 810 (22.9) <0.001CHA2DS2VASc score <0.001 0, n (%) 3009 (16.1) 430 (12.1) 1, n (%) 8623 (46.3) 1039 (29.3) 2, n (%) 4546 (24.4) 1067 (30.1) 3, n (%) 1755 (9.4) 623 (17.6) ≥4, n (%) 704 (3.8) 383 (10.8)

Table 1. Baseline characteristics of the study population

AF = atrial fibrillation, BMI = body mass index, CABG = coronary artery bypass graft, CVD = cardiovascular disease, PCI = percutaneous coronary intervention, SD = standard deviation. Statistically significant P values are in bold.

17

Events CHA2DS2VASc Events (n)Person-years (n)

Incidence rate % (95% CI)

Events (n)Person-years (n)

Incidence rate % (95% CI)

0 166 44396 0.37 (0.32-0.44) 55 6345 0.87 (0.65-1.13)1 402 124323 0.32 (0.29-0.36) 171 14388 1.19 (1.02-1.38)2 369 61996 0.60 (0.54-0.66) 215 13969 1.54 (1.34-1.76)3 216 21008 1.03 (0.90-1.17) 145 7288 1.99 (1.68-2.34)

≥4 84 7431 1.13 (0.90-1.40) 76 3908 1.94 (1.53-2.43)0 133 44580 0.30 (0.25-0.35) 49 6356 0.77 (0.57-1.02)1 310 124754 0.25 (0.22-0.28) 150 14440 1.04 (0.88-1.22)2 302 62256 0.49 (0.43-0.54) 199 14009 1.42 (1.23-1.63)3 168 21175 0.79 (0.68-0.92) 130 7350 1.77 (1.48-2.10)

≥4 63 7479 0.84 (0.65-1.08) 68 3927 1.73 (1.34-2.20)0 261 43892 0.59 (0.52-0.67) 66 6298 1.05 (0.81-1.33)1 518 123948 0.42 (0.38-0.46) 144 14561 0.99 (0.83-1.16)2 416 62207 0.67 (0.61-0.74) 194 14140 1.37 (1.19-1.58)3 251 20979 1.20 (1.05-1.35) 113 7507 1.51 (1.24-1.81)

≥4 119 7457 1.60 (1.32-1.91) 80 3957 2.02 (1.60-2.52)0 103 45396 0.23 (0.19-0.28) 37 6654 0.56 (0.39-0.77)1 316 126498 0.25 (0.22-0.28) 137 15221 0.90 (0.76-1.06)2 377 63869 0.59 (0.53-0.65) 178 15011 1.19 (1.02-1.37)3 254 21876 1.16 (1.02-1.31) 184 7965 2.31 (1.99-2.67)

≥4 160 7774 2.06 (1.75-2.40) 143 4229 3.38 (2.85-3.98)0 456 45396 1.00 (0.91-1.10) 87 6654 1.31 (1.05-1.61)1 1410 126498 1.11 (1.06-1.17) 336 15221 2.21 (1.98-2.46)2 1356 63869 2.12 (2.01-2.24) 450 15011 3.00 (2.73-3.29)3 794 21876 3.63 (3.38-3.89) 339 7965 4.26 (3.82-4.73)

≥4 395 7774 5.08 (4.59-5.61) 276 4229 6.53 (5.78-7.34)

Table 2. Crude Incidence Rates at 20 Years of Follow-up

All-cause mortality

NO AF AF

All stroke

Ischaemic stroke

Coronary event

Cardiovascular mortality

18

Variable Hazard Ratio % (95% CI)

P

Age at event 0.90 (0.88-0.91) <0.001Dyslipidemia 1.39 (1.01-1.79) 0.012AF, any type 3.03 (2.72-3.38) <0.001Current smoker (vs. non-smoker) 1.30 (1.15-1.48) <0.001Antithrombotic drug use 2.96 (2.05-3.43) <0.001CHA2DS2VASc ≥2 2.01 (1.84-2.36) <0.001

Statistically significant P values are in bold.

Table 3. Multivariate Cox regression for Ischaemic Stroke

19

FIGURE LEGENDS

Figure 1. Flow chart summarising the selection process of study population.

Figure 2. Cumulative probability of ischaemic stroke-free survival in subjects without (AF -ve,

panel A) and with AF (AF +ve, panel B) by CHA2DS2VASc strata. Dashed blue line represents

subjects with CHA2DS2VASc score=0, dashed red CHA2DS2VASc score=1, dashed green

CHA2DS2VASc score=2, dashed light-blue CHA2DS2VASc score=3, dashed purple CHA2DS2VASc

≥4. Bonferroni-corrected Log-rank P-values for pairwise comparisons between different

CHA2DS2VASc strata are also displayed within the graph.

Figure 3. Cumulative probability of ischaemic stroke-free survival in subjects without AF (AF -ve)

and a CHA2DS2VASc score ≥4 (dashed blue line) and in subjects with AF (AF +ve) and a

CHA2DS2VASc score = 2 (dashed red line) at 20-years follow-up (panel A) and 10-years follow-up

(panel B).

Figure 4. Cumulative probability of AF-free survival by CHA2DS2VASc strata. Dashed blue line

represents subjects with CHA2DS2VASc score=0, dashed red CHA2DS2VASc score=1, dashed

green line CHA2DS2VASc score=2, dashed light-blue CHA2DS2VASc score=3, dashed purple line

CHA2DS2VASc≥4. Bonferroni-corrected Log-rank P-values for pairwise comparisons between

different CHA2DS2VASc strata are also displayed within the graph.

20