Burden of atrial fibrillation and poor rate controldetected by continuous monitoring and the risk forheart failure hospitalizationShantanu Sarkar, PhD, a Jodi Koehler, MS, a George H. Crossley, MD, b W. H. Wilson Tang, MD, c

William T. Abraham, MD, d Eduardo N. Warman, PhD, a and David J. Whellan, MDe Mounds View, MN;Nashville, TN; Cleveland, and Columbus, OH; and Philadelphia, PA

Introduction Atrial fibrillation (AF) on electrocardiogram has been identified as a risk factor for hospitalizations inpatients with heart failure (HF). We investigated whether continuous AF monitoring can identify when patients with HF are atrisk for hospitalization.

Methods In this retrospective analysis of data from 4 studies enrolling patients with HF with cardiac resynchronizationtherapy defibrillator devices with ≥90 days of follow-up (n = 1561), patients were identified as having AF if they had ≥1 day ofN5 minutes of AF and N1 hour of total AF during entire follow-up. In patients with AF, device recorded AF burden (AFb) andventricular rate during AF (VRAF) over the last 30 days was classified on a monthly basis into 3 evaluation groups: (1) ≥1 dayof high burden of paroxysmal AF (≥6 hours) or persistent AF (all 30 days with AFb N23 hours) with poor rate control (VRAF N90beats/min), (2) ≥1 day of high burden of paroxysmal AF with good rate control (VRAF ≤ 90 beats/min), and (3) no days withhigh burden of AF (AFb b6 hours) or persistent AF with good rate control. Each group was compared with monthly evaluationsin patients without AF using an Anderson-Gill model for occurrence of HF hospitalizations in the next 30 days.

Results Patients with AF (n = 519, 33%) have a greater risk (hazard ratio [HR] 2.0, P b .001) for impending HFhospitalizations during entire follow-up compared with patients with no AF. One day of high burden of paroxysmal AF withgood rate control in the last 30 days increases risk for HF hospitalization in the next 30 days (HR 3.4, P b .001). The riskincreases further (HR 5.9, P b .001) with 1 day of poor rate control during persistent AF or high burden paroxysmal AF in last30 days.

Conclusion Evaluation of AFb and rate control information on a monthly basis can identify patients at risk for HFhospitalization in the next 30 days. (Am Heart J 2012;164:616-24.)

BackgroundIn the United States, N2.2 million people have atrial

fibrillation (AF) and N5.7 million have heart failure (HF).1

In patients hospitalized with AF, the most commonprimary diagnosis is HF.1 In patients with HF, AF is themost common arrhythmia.2-4 The Euro Heart Surveysshowed that HF is present in 34% of patients with AF3 andAF is present in 42% of patients with HF.4 In patients with

rom the aMedtronic Inc, Mounds View, MN, bSt Thomas Research Institute andniversity of Tennessee College of Medicine, Nashville, TN, cThe Cleveland Clinic,leveland, OH, dThe Ohio State University, Columbus, OH, and eThomas Jeffersonniversity, Philadelphia, PA.ubmitted April 10, 2012; accepted June 22, 2012.eprint requests: Shantanu Sarkar, PhD, Senior Principal Scientist, Medtronic Incorporated,200 Coral Sea St, MVN41, Mound View, MN 55112.-mail: shantanu.sarkar@medtronic.com002-8703/$ - see front matter2012, Mosby, Inc. All rights reserved.

FUCUSR8E0©

http://dx.doi.org/10.1016/j.ahj.2012.06.020

AF or HF, subsequent development of the other conditionwas associated with increased mortality.2 Atrial fibrilla-tion is as common in patients with HF with low ejectionfraction (EF), as it is in patients with HF with preservedEF.5-7 Patients with HF with AF and low EF have higherabsolute risk for cardiovascular death or HF hospitaliza-tion, and patients with HF with AF and preserved EF havea higher relative risk comparedwith patients without AF.7

Implanted devices such as cardiac resynchronizationtherapy defibrillator (CRT-D) devices, implantable cardi-overter defibrillators (ICD), pacemakers, and implantableloop recorders (ILRs) continuously monitor AF burden(AFb) (including atrial tachycardias) and ventricular rateduring AF (VRAF) (Figure 1). These devices have beenshown to have a high accuracy for detecting AF usingsensing leads in the atrium in CRT-D/ICD devices8 andusing subcutaneous electrodes in ILRs.9 Although asnapshot measurement of AF with electrocardiogramshas been identified as a significant independent predictor

Figure 1

The atrial tachyarrhythmia burden trends in a Medtronic ICD/CRT-D/pacemaker/ILR device.

Sarkar et al 617American Heart JournalVolume 164, Number 4

of clinical events, including HF hospitalizations,10 previ-ous studies have shown that intermittent monitoringusing electrocardiograms and Holter recordings do notdetect atrial arrhythmias as well as implantable continu-ous monitoring.11 Recently, it was shown that implant-able device–detected AF is associated with increased riskfor stroke12,13 and death or HF hospitalization.14 Thisstudy investigates whether dynamic assessment of atrialarrhythmia burden and VRAF at regular monthly follow-up can be used to identify periods when patients are atincreased risk for HF hospitalization in the next 30 days.

MethodsData set and event definitionsThe retrospective analysis included data available from the

PARTNERS-HF15 (N = 699 patients), FAST16 (N = 146 patients),OFISSER17 (N = 323 patients), Case Study Registry (N = 80patients), and CONNECT18 (N = 313 patients) studies.PARTNERS-HF was a prospective observational study enrollingpatients with CRT-D devices with EF ≤35%, New York HeartAssociation (NYHA) class III or IV, and QRS duration ≥130 ms.FAST was a prospective double-blinded observational study inpatients with CRT-D or ICD devices with EF ≤35% and NYHAclass III or IV. The OFISSER and Case Study Registry includedpatients with CRT-D devices where data were collected using aretrospective review. The CONNECT study was a prospectiverandomized study including patients with ICD or CRT-D devicesthat were willing to transmit data remotely. The control armpatients from the CONNECT study, who were not being activelymonitored based on AF diagnostic alerts, were included for theanalysis. Patients with permanent AF were excluded in theCONNECT and PARTNERS-HF studies. Only patients with CRT-Ddevices, that is, patients with HF with reduced left ventricularfunction and at least 90 days of follow-up data were included inthe data cohort of 1,561 patients. The method for detecting AF issame in all the devices included for the data analysis.Emergency department visits with intravenous diuretic

administration and HF hospitalizations were used as the HFend point in the data analysis. Each cardiovascular hospitaliza-tion was carefully adjudicated for signs and symptoms of HF,

which included administration of intravenous or oral diureticduring the hospitalization. Death was not used as an end point inthe data analysis.

Patient cohort and monthly diagnostic evaluationgroup definitions

Atrial fibrillation patient cohort. An AF patientcohort was formed using the patients with HFwith AF (Figure 2).A patient is defined to have any AF if the patient had devicedetected AFb N5minutes on at least 1 day and a total AFb N1 hourduring the entire follow-up. These limits were predefined toavoid patients with falsely detected AF and patients without asignificant amount of AF being included in the AF patient cohort.The patients who did not satisfy the criteria for the AF patientcohort were in the “no-AF” patient cohort.

Monthly diagnostic evaluation groups in the AFpatient cohort. In the monthly evaluation scheme shown inFigure 3, every 30 days device recorded AFb and VRAF wereevaluated for the last 30 days, and the monthly status for patientsin the AF cohort was classified into 3 groups (Figure 2): (1) AFbwith rapid ventricular rate (AFb + RVR) group, (2) AFb withnormal ventricular rate (AFb + NVR) group, and (3) “other AF”group consisting of the evaluations that do not get categorized ingroup 1 or 2. Each patient in the AF cohort may be part ofdifferent groups at different evaluations depending on their AFband VRAF status in the 30 days before each diagnosticevaluation. To determine the criteria for AFb in the AFb +RVR and AFb + NVR groups, 4 different thresholds wereinvestigated based on AFb in last 30 days: (1) no days with AFb≥6 hours, (2) 1 to 6 days with AFb ≥6 hours, (3) ≥7 days withAFb ≥6 hours, and (4) all 30 days with AFb ≥23 hours (definedas persistent AF). The AFb thresholds that had higher relativerisks were grouped into the AFb + RVR or the AFb + NVR groupsdepending on the rapid VRAF status at the monthly evaluation,defined as at least 1 day with average VRAF N90 beats/min in thelast 30 days. The thresholds, such as 90 beats/min for VRAF and6 hours and 23 hours for AFb, are thresholds already in use fordiagnostic observations in Medtronic CRT-D devices and havebeen identified as predictive thresholds in previous studies.15

Secondary patient cohorts. With the intention ofselecting patients with higher baseline risk, the AF patientcohort was further stratified into 2 subcohorts: (1) significant

Figure 2

The basic schematic of how monthly diagnostic evaluation groups were derived from the AF and “no-AF” patient cohorts.

Figure 3

Performance evaluation schematic to evaluate whether AF diagnosticscan identify the periods when patients with HF are at higher risk forHF hospitalizations.

618 Sarkar et alAmerican Heart Journal

October 2012

AFb cohort consisted of patients with ≥1 day of AFb ≥6 hoursduring the entire follow-up (ie ≥1 monthly evaluation satisfyingthe AF with rapid ventricular rate or AFb + NVR or persistent AFcriteria), (2) significant AFb and poor rate control cohortconsisted of patients with ≥1 day of AFb ≥6 hours and averageVRAF ≥90 beats/min during the entire follow-up (ie, ≥1monthly evaluation satisfying the AF with rapid ventricularrate criteria). The significant AFb and poor rate control patientcohort is a subset of the significant AFb patient cohort, which isa subset of the AF patient cohort.

Statistical analysisThe baseline variables between the AF patient cohort and the

“no-AF” patient cohort were compared using the Student t test

for continuous variables and the χ2 test for categorical variables.Time to first HF hospitalization over the entire follow-up wascompared between patient cohorts using the Cox proportionalhazards model after adjusting for baseline variables (age, gender,NYHA, history of coronary artery disease, and myocardialinfarction [MI]) and baseline medications (angiotensin-convertingenzyme inhibitor/angiotensin receptor blocker [ACE-I/ARB],diuretics, β-blockers, antiarrhythmic drugs). The model was notadjusted for EF or QRS width as the actual value of EF and QRSwidth was not collected in all studies; whether EF was b35% wasonly collected.Each monthly evaluation (Figure 3) included: (1) a 30-day

retrospective look at AF diagnostic data to ascertain patientstatus into the diagnostic evaluation groups and (2) aprospective assessment for the first HF hospitalization in thenext 30 days. The evaluation scheme in Figure 3 models themonthly remote device interrogation used in clinical practiceand was used for evaluation in earlier studies.15 A monthlyevaluation was included only if there was N30 days of devicedata and clinical follow-up following the diagnostic evaluation.Patients who died during the studies were included in theanalysis, but deaths were excluded as an end point from theanalysis using the requirement of 30 day of device follow-upafter evaluation. Monthly evaluations in patients with “no-AF”were considered as the reference group. The 3 monthlydiagnostic evaluation groups in the AF patient cohort werecompared with the reference group for time to first HFhospitalization in the next 30 days using the Anderson-Gillmodel, an extension of the Cox proportional hazards model thataccounts for multiple evaluations in patients. The model wasadjusted for baseline variables (age, gender, NYHA, history ofcoronary artery disease, and MI) and baseline medications (ACE-I/ARB, diuretics, β-blockers, antiarrhythmic drugs).A secondary analysis was performed to investigate whether

dynamic assessment of AF diagnostics can identify periods when

Table. Comparison of baseline demographics in patients withand without AF in patients with HF with left ventricular systolicdysfunction

Total(n = 1561)

Patientswith AF(n = 519)

Patientswithout AF(n = 1042) P

Mean age (SD) 69 (11) 70 (10) 68 (11) b.001Male 69% 76% 65% b.001Mean EF (SD) 25 (8) 25 (8) 25 (8) .70EF ≤35% 98% 97% 98% .17NYHAI 2% 1% 2% .46II 12% 12% 12%III 82% 80% 82%IV 5% 6% 4%

CAD 66% 69% 65% .08HTN 71% 72% 71% .53Diabetes 42% 44% 42% .38MI 43% 41% 44% .23History of AF 27% 50% 17% b.001Baseline medicationsACE-I/ARB 78% 79% 78% .67β-Blockers 89% 86% 90% .01Diuretics 82% 84% 81% .09Digoxin 34% 40% 32% .001Aldosterone

receptor blocker32% 31% 32% .68

Antiarrhythmic 21% 29% 16% b.001Antithrombotic 82% 86% 79% .001Warfarin 33% 49% 25% b.001

Abbreviations: CAD, coronary artery disease; HTN, hypertension.

Figure 4

aplan-Meier curves for time to first HF hospitalization in patientsith and without AF in patients with HF with left ventricularystolic dysfunction.

Sarkar et al 619American Heart JournalVolume 164, Number 4

a patient is at higher risk across different patient cohorts withhigher baseline risk. Considering only patients in each of the 3patient cohorts (1) AF patient cohort, (2) significant AFb patientcohort, and (3) significant AFb and poor rate control patientcohort, the 3 monthly evaluation groups were created andcompared separately for each subset of patients using theevaluation group definitions defined previously. Because thepatients with “no-AF” are not a part of any of the patient cohorts,the “other AF” monthly evaluation group within each cohortwas used as the reference group instead of using the evaluationsin patients with “no-AF” as reference for this secondary analysis.

The clinical studies from which data were pooled to conductthis investigation were sponsored by Medtronic Incorporated.No extramural funding was used to support this work. Theauthors are solely responsible for the design and conduct of thisstudy, all study analyses, the drafting and editing of themanuscript, and its final contents.

ResultsA total of 1,561 patients with an average follow-up

duration of 373 ± 146 days were used for this analysis. Atotal of 326 HF hospitalizations were identified in 207patients (13%) providing an event rate of 0.2 per patient-year of monitoring. The baseline characteristics of thepatients in the study are shown in the Table. Patients inthe AF cohort were older and more likely to be males

Kws

compared with patients with “no-AF.” No significantdifferences were observed for ejection fraction, history ofcoronary artery disease, hypertension, diabetes, or MI.Patients with AF were more likely to be on digoxin,antiarrhythmic, and warfarin and less likely to be on β-blockers. No significant differences were observed forACE-I/ARB, diuretic, or aldosterone receptor blockerusage between the groups. Patients in the AF cohort had agreater risk for HF hospitalization during the entirefollow-up (hazard ratio [HR] 2.0, 95% CI 1.5-2.7, P b .001)compared with patients with “no-AF” (Figure 4).The event rates and adjusted HRs for the 4 different AFb

thresholds at rapid and normal ventricular rates investi-gated in the monthly evaluation model are presented inFigure 5. The event rate is expressed as a percentage ofmonthly evaluations that were followed by an HFhospitalization in the next 30 days. For evaluations withnormal VRAF, risk for HF hospitalization increased withthe amount of AFb (HR 2.1-3.9) except for persistent AF.Furthermore, AF ≥6 hours combined with poor ratecontrol increased that risk (HR 5.3-5.8). Interestingly,persistent AF at all times does not increase the risk unlessit was associated with rapid ventricular rate (HR 6.7). Inthe AF patient cohort, the event rate following a 30-dayperiod with no AF was 2.2%, which is similar to the eventrate following a 30-day period with ≥1 day with AF b6hours, but both are greater than the event rate in the “no-AF” patient cohort as shown in Figure 5. Based on theresults in Figure 5, the AFb + RVR monthly evaluationgroup composed of evaluations that have at least 1 daywith both AFb≥6 hours and average VRAF N90 beats/mincriteria met in last 30 days (Figure 6). Similarly, the AFb +NVR group consisted of evaluations that have at least 1day with AFb≥6 hours and do not meet the criteria of the

Figure 5

Heart failure hospitalization rate as percentage of monthly evaluations for the different AFb and rate control groups. Adjusted HRs for comparisonsbetween different AFb thresholds and the reference are shown in the figure.

620 Sarkar et alAmerican Heart Journal

October 2012

AFb + RVR group but excludes evaluations that have all30 days with AFb ≥23 hours because persistent AF withgood rate control does not have increased risk comparedwith evaluations in patients with “no-AF” (Figure 6).Figure 7 shows the Kaplan-Meier plot for time to first

HF hospitalization in the 30 days following monthlydiagnostic evaluation and the adjusted HRs for compar-ison between the 3 evaluation groups in the AF patientcohort and the reference group. A total of 228 monthlyevaluations (1.5%) were followed by an HF hospitaliza-tion in the next 30 days. Of the 315 monthly evaluationswhen the AFb + RVR criteria were satisfied, 20 (6.3%)were followed by a HF hospitalization in the next 30 days.The AFb + RVR criteria were satisfied in at least 1 monthlyevaluation in 148 patients (9.5%). Monthly diagnosticevaluations with AFb + RVR criteria met were 5.9 times(HR 5.9, 95% CI 3.5-10.0, P b .001) more likely to have anHF hospitalization in the next 30 days compared withmonthly evaluations in patients with “no-AF.” The AFb +RVR or the AFb + NVR criteria were satisfied in 969monthly evaluations in 300 patients (19.2%) with 41 HFhospitalizations in the next 30 days (event rate of 4.2%).Figure 8 shows that the HRs for event rate comparisons

between the 3 monthly evaluation groups are very similarin different patient cohorts. As an example, patients in thesignificant AFb and poor rate control cohort were at ahigher baseline risk for HF hospitalizations over the entirefollow-up duration compared with patients with “no-AF”

(HR 2.5, 95% CI 1.7-3.7, P b .001). In addition, monthlyevaluations satisfying the criteria for the AFb + NVRevaluation groupwithin this patient cohortwere 3.8 timesmore likely to be hospitalized for HF in the next 30 dayscompared with monthly evaluations satisfying the criteriaof the “other AF” group within this patient cohort.Beyond risk-stratifying patients on a static basis based

on whether they have AF (Figure 4), the results in Figures5 and 7 show that the amount of AF, type of AF, andchange in rate control status evaluated on a dynamic basisprovide incremental information that can identify month-ly periods when patients are at higher risk for HFhospitalizations. Furthermore, Figure 8 shows that theresults hold true irrespective of the choice of patientcohorts with different baseline risk for HF hospitalization.

DiscussionThe study presented a retrospective data analysis to

evaluate the ability of continuous AF monitoring topredict future HF hospitalizations. One day with highburden (N6 hours) of paroxysmal AF with good ratecontrol in last 30 days (AFb + NVR group) increases therisk for HF hospitalization in the next 30 days. The riskincreases further if there is 1 day of poor rate controlduring persistent AF or high burden paroxysmal AF (AFb+ RVR group). Thus, a high burden of paroxysmal AF (N6hours) with good or poor rate control is a risk factor for

Figure 7

Kaplan-Meier curves for time to first HF hospitalization (HFH) aftermonthly diagnostic evaluation for the different diagnostic evaluationgroups. Adjusted HRs for comparisons between the 3 evaluationgroups in the AF patient cohort and the reference group composing ofevaluations in the patients with “no-AF” are shown in the figure.

Figure 6

Flow diagram of AF diagnostic criteria used to determine the different evaluation groups at each monthly evaluation.

Sarkar et al 621American Heart JournalVolume 164, Number 4

HF hospitalization; however, persistent AF is a risk factoronly in conjunction with poor rate control. Earlier studieshave established that knowing that the patient has AFidentifies which patient needs more clinical attention.This study further shows that information on whenpatient is having AF, amount of AF, type of AF, and thechanges in rate control status during AF may provide anopportunity to proactively provide more clinical atten-tion to selective patients in a timely manner to reduceHF hospitalizations.Atrial fibrillation burden with rapid ventricular rate is

one of the known causes for loss in cardiac resynchroniza-tion therapy (CRT) pacing. The primary analysis did notinclude adjustment for CRT pacing in the model. Thedeterministic cause-and-effect relationship between AFb+ RVR and loss of CRT pacing would confound theinterpretation of the results after adjustment for loss ofCRT pacing because the adjustment would underestimatethe relative risk of an HF hospitalization due to AFb + RVR.When adjusted for loss of CRT pacing in the model, theAFb + RVR and the AFb +NVR evaluation groupswere stillmore likely to have HF hospitalization in the next 30 dayscompared with the reference group (HR 4.0 and 3.0,respectively, P b .001). Thus, AF diagnostics provideinformation that is independent to that provided by a lossof CRT pacing diagnosticwith respect to identifyingwhenpatients are at risk for HF hospitalizations.

Combined treatment strategies in patients with AF andHF have been evaluated in the last decade.6,19 Most studieshave shown that rhythm control strategy based on

Figure 8

Heart failure hospitalization rate as percentage of monthly evaluations for the different diagnostic evaluation groups in AF cohort (A), significantAFb cohort (B), and significant AFb and poor rate control cohort (C).

622 Sarkar et alAmerican Heart Journal

October 2012

pharmacologic treatments and electrical cardioversiondoes not reduce mortality compared with pharmacologicrate control strategies.20,21 Lenient rate control has beenshown to be as effective as strict rate control in patientswith permanent AF.22 However, the study included veryfew patients with NYHA-III HF symptoms or previous HFhospitalizations or paroxysmal AF, the main characteristicsof the patient cohort in this study. Furthermore, intermit-tent monitoring using Holter underestimates incidences ofpoor rate control in permanent AF patients.23 Continuouslong-term monitoring of AF using implantable devices withremote access and wireless alerting capabilities providesdynamic assessment of AF type and rhythm and ratecontrol status, thus providing the opportunity to optimizepharmacologic treatment strategies24-28 for patients withHF with AF in a timely manner. Interestingly, very fewstudies have evaluated the possibility of dynamic optimi-zation of HF therapies, such as diuretics, β-blockers, ACE-I/ARB, and CRT pacing, based on AF diagnostic information.Rate control medications may worsen HF symptoms in theshort term if patient is already hypervolemic. When the AFdiagnostic identifies risk of HF, one strategy could be toevaluate fluid status and HF signs and symptoms first. Iffluid overload is corroborated, then it may suggesttherapeutic actions for controlling fluid status, such asdiuretics, in the short term. Once the patient is euvolemic,poor rate control during AF can be addressed by

optimizing pharmacologic rate control therapy, such asβ-blockers, to prevent future HF events. Furthermore,pulmonary vein isolation29 or atrioventricular nodeablation30 can also be considered for long-term reductionof AFb and poor rate control in patients with HF. Whethertherapeutic interventions based on continuous AF diag-nostic information improves outcomes in patients with AFand HF need carefully designed prospective evaluation toshow the safety and efficacy of such a disease managementparadigm. Several randomized controlled studies formanagement of patients with HF based on diagnosticinformation have yielded inconsistent results.31-35

LimitationsThe retrospective analysis was done by pooling data

collected in 4 previous studies to increase sample size forthe data cohort. Dynamic assessment of clinical diagnos-tic data such as serial measurement of weight, bloodpressure, and B-type natriuretic peptide was not per-formed in all of these studies. Thus, the statistical analysisis limited to adjustment of baseline clinical variables only.The incremental value of AF diagnostic over clinicaldiagnostic measurements cannot be established. It ishypothesized that dynamic assessment of AF providessupplementary information that should aid HF diagnosticassessment. The overall goal would be to combine the

Sarkar et al 623American Heart JournalVolume 164, Number 4

different diagnostic measurements related to HF toimprove the efficacy of disease management programs.

ConclusionThe current study identifies thresholds for creating

different AFb and rate control groups and shows thatevaluation of AFb and rate control information on amonthly basis can identify patients at risk for HFhospitalization in the next 30 days. Future studies areneeded to validate the chosen threshold for categorizingdifferent risk groups in a broader HF patient populationand to evaluate whether timely intervention based on AFdiagnostics can reduce cardiovascular hospitalizations inpatients with HF and AF.

References1. Roger VL, et al. Heart Disease and Stroke Statistics—2011 update: a

report from the American Heart Association. Circulation 2011;123:e18-e209.

2. Wang TJ, Larson MG, Levy D, et al. Temporal relations of atrialfibrillation and congestive heart failure and their joint influence onmortality: the Framingham Heart Study. Circulation 2003;107:2920-5.

3. Nieuwlaat R, Capucci A, Camm AJ, et al. Atrial fibrillationmanagement: a prospective survey in ESC member countries: theEuro Heart Survey on Atrial Fibrillation. Eur Heart J 2005;26:2422-34.

4. Cleland JG, Swedberg K, Follath F, et al. The EuroHeart Failure surveyprogramme— a survey on the quality of care among patients withheart failure in Europe. Part 1: patient characteristics and diagnosis.Eur Heart J 2003;24:442-63.

5. Lenzen MJ, Scholte op Reimer WJ, Boersma E, et al. Differencesbetween patients with a preserved and a depressed left ventricularfunction: a report from the EuroHeart Failure Survey. Eur Heart J2004;25:1214-20.

6. Nieuwlaat R, Eurlings LW, Cleland JG, et al. Atrial fibrillation andheart failure in cardiology practice: reciprocal impact and combinedmanagement from the perspective of atrial fibrillation: results of theEuro Heart Survey on atrial fibrillation. J Am Coll Cardiol 2009;53:1690-8.

7. Olsson LG, Swedberg K, Ducharme A, et al. Atrial fibrillation and riskof clinical events in chronic heart failure with and without leftventricular systolic dysfunction: results from the Candesartan in Heartfailure-Assessment of Reduction in Mortality and morbidity (CHARM)program. J Am Coll Cardiol 2006;47:1997-2004.

8. Swerdlow CD, Schsls W, Dijkman B, et al. Detection of atrialfibrillation and flutter by a dual-chamber implantablecardioverter-defibrillator. Circulation 2000;101:878-85.

9. Hindricks G, Pokushalov E, Urban L, et al. Performance of a newleadless implantable cardiac monitor in detecting and quantifyingatrial fibrillation: results of the XPECT trial. Circ ArrhythmElectrophysiol 2010;3:141-7.

10. Dries DL, Exner DV, Gersh BJ, et al. Atrial fibrillation is associatedwith an increased risk for mortality and heart failure progression inpatients with asymptomatic and symptomatic left ventricular systolicdysfunction: a retrospective analysis of the SOLVD trials. Studies ofLeft Ventricular Dysfunction. J Am Coll Cardiol 1998;32:695-703.

11. Ziegler P, Koehler J, Mehra R. Comparison of continuous versusintermittent monitoring of atrial arrhythmias. Heart Rhythm 2006;3:1445-52.

12. Glotzer TV, Daoud EG, Wyse DG, et al. The relationship betweendaily atrial tachyarrhythmia burden from implantable device di-agnostics and stroke risk: the TRENDS study. Circ ArrhythmElectrophysiol 2009;2:474-80.

13. Healey JS, Connolly SJ, Gold MR, et al. Subclinical atrial fibrillationand the risk of stroke. N Engl J Med 2012;366:120-9.

14. Santini M, Gasparini M, Landolina M, et al. Device-detected atrialtachyarrhythmias predict adverse outcome in real-world patients withimplantable biventricular defibrillators. J Am Coll Cardiol 2011;57:167-72.

15. Whellan DJ, Ousdigian KT, Al-Khatib SM, et al. Combined heartfailure device diagnostics identify patients at higher risk of subsequentheart failure hospitalizations: results from PARTNERS HF (Program toAccess and Review Trending Information and Evaluate Correlation toSymptoms in Patients With Heart Failure) study. J Am Coll Cardiol2010;55:1803-10.

16. Abraham WT, Compton S, Haas G, et al. Superior performance ofintrathoracic impedance-derived fluid index versus daily weightmonitoring in heart failure patients: results of the Fluid AccumulationStatus Trial (FAST). Congest Heart Fail 2011;17:51-5.

17. Small RS,WickemeyerW, Germany R, et al. Changes in intrathoracicimpedance are associated with subsequent risk of hospitalizations foracute decompensated heart failure: clinical utility of implanted devicemonitoring without a patient alert. J Card Fail 2009;15:475-81.

18. Crossley GH, Boyle A, Vitense H, et al. CONNECT Investigators. TheCONNECT (Clinical Evaluation of Remote Notification to ReduceTime to Clinical Decision) Trial The Value of Wireless RemoteMonitoring With Automatic Clinician Alerts. J Am Coll Cardiol 2011;57:1181-9.

19. Piccini JP, Hernandez AF, Zhao X, et al. Quality of care for atrialfibrillation among patients hospitalized for heart failure. J Am CollCardiol 2009;54:1280-9.

20. Roy D, Talajic M, Nattel S, et al. Rhythm control versus rate controlfor atrial fibrillation and heart failure. N Engl J Med 2008;358:2667-77.

21. Talajic M, Khairy P, Levesque S, et al. Maintenance of sinus rhythmand survival in patients with heart failure and atrial fibrillation. J AmColl Cardiol 2010;55:1796-802.

22. Van Gelder IC, Groenveld HF, Crijns HJ, et al. Lenient versus strict ratecontrol in patients with atrial fibrillation. N Engl J Med 2010;362:1363-73.

23. Ziegler PD, Koehler JL, Verma A. Continuous versus intermittentmonitoring of ventricular rate in patients with permanent atrialfibrillation. Pacing Clin Electrophysiol 2012;35:598-604.

24. Camm AJ, et al. Guidelines for the management of atrial fibrillation:the Task Force for the Management of Atrial Fibrillation of theEuropean Society of Cardiology (ESC). Europace 2010;12:1360-420.

25. Hohnloser SH, Crijns HJ, van Eickels M, et al. Dronedarone in patientswith congestive heart failure: insights from ATHENA. Eur Heart J2010;31:1717-21.

26. Connolly SJ, Camm AJ, Halperin JL, et al. Dronedarone in high-riskpermanent atrial fibrillation. N Engl J Med 2011;365(24):2268-76.

27. Fauchier L, Grimard C, Pierre B, et al. Comparison of beta blockerand digoxin alone and in combination for management of patientswith atrial fibrillation and heart failure. Am J Cardiol 2009;103:248-54.

28. Wallentin L, Yusuf S, Ezekowitz MD, et al. Efficacy and safety ofdabigatran compared with warfarin at different levels of international

624 Sarkar et alAmerican Heart Journal

October 2012

normalised ratio control for stroke prevention in atrial fibrillation: ananalysis of the RE-LY trial. Lancet 2010;376:975-83.

29. Hsu LF, Jaïs P, Sanders P, et al. Catheter ablation for atrialfibrillation in congestive heart failure. N Engl J Med 2004;351:2373-83.

30. Doshi RN, Daoud EG, Fellows C, et al. Left ventricular-based cardiacstimulation post AV nodal ablation evaluation (the PAVE study).J CardiovascElectrophysiol 2005;16:1160-5.

31. Kimmelstiel C, Levine D, Perry K, et al. Randomized, controlledevaluation of short- and long-term benefits of heart failure diseasemanagement within a diverse provider network: the SPAN-CHF trial.Circulation 2004;110:1450-5.

32. Goldberg LR, Piette JD, Walsh MN, et al. Randomized trial of a dailyelectronic home monitoring system in patients with advanced heart

Receive tables of cont

To receive the tables of contents by e-mail, sig

http://www.ahjo

Choose E-mail Notification

Simply type your e-mail address in the box an

Alternatively, you may send on e-mail messag

majordomo@mo

Leave the subject line blank, and type the follo

subscribe ah

You will receive an e-mail to confirm that you

Note that TOC e-mails will be sent when a n

failure: the Weight Monitoring in Heart Failure (WHARF) trial. AmHeart J 2003;146:705-12.

33. Cleland JG, Louis AA, Rigby AS, et al. Noninvasive hometelemonitoring for patients with heart failure at high risk of recurrentadmission and death: the Trans-European Network-Home-CareManagement System (TEN-HMS) study. J Am Coll Cardiol 2005;45:1654-64.

34. Bourge RC, Abraham WT, Adamson PB, et al. Randomizedcontrolled trial of an implantable continuous hemodynamic monitor inpatients with advanced heart failure: the COMPASS-HF study. J AmColl Cardiol 2008;51:1073-9.

35. Abraham WT, Adamson PB, Bourge RC, et al. Wireless pulmonaryartery haemodynamic monitoring in chronic heart failure: arandomised controlled trial. Lancet 2011;377:658-66.

ents by e-mail

n up through our Web site at

nline.com

d click on the Subscribe button

e to

sby.com

wing as the body of your message:

j_toc

have been added to the mailing list.

ew issue is posted to the Web site.