J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 3 , N O . 5 , 2 0 1 7 Pathak et al.M A Y 2 0 1 7 : 4 3 6 – 4 7 Cost-Effectiveness of Risk Factor Management in AF
437
AB BR E V I A T I O N S
AND ACRONYM S
AF = atrial fibrillation
AFSS = Atrial Fibrillation
Severity Scale
CI = confidence interval
HR = hazard ratio
ICER = incremental cost-
effectiveness ratios
QALY = quality-adjusted life
years
RFM = risk factor management
R ecently reported epidemiological dataconfirm the emergence of atrial fibrillation(AF) as a global epidemic (1). This has signif-
icant and progressive impact on health care costsbecause of its association with increased cardiovascu-lar morbidity, reduced quality of life, stroke, andmortality (2–4). The incremental cost of AF in theUnited States is estimated to range between $6.0billion and $26.0 billion per year (5). Hospitalization,increased medication use, and procedural require-ments constitute the major contributors to the totaltreatment cost of patients with AF (6–9). Ageing pop-ulations are an important contributor to the growingburden of AF. Recent data have also implicated theincreasing prevalence of risk factors, such as obesity,hypertension, diabetes mellitus, and obstructivesleep apnea (10,11). Therefore, there is an urgentneed for improved and cost-effective primary andsecondary prevention strategies to reduce the impactof this enormous health burden.
SEE PAGE 448
Sinus rhythm is associated with better quality oflife (12). Detrimental effects of antiarrhythmic agentsoffset the benefit from sinus rhythm maintenance(13,14). Catheter ablation of AF has evolved as aneffective therapy for drug-refractory symptomatic AF(15). However, it is resource intensive and has sig-nificant upfront costs. Furthermore, reports of long-term outcomes demonstrate attrition in successwith time (16–18). The cost-effectiveness of AF abla-tion is greatly influenced by the number of proced-ures, their success rate, and procedural complications(19,20). Studies have associated cardiac risk factorswith the more frequent recurrence of AF, increased
manuscript; or the decision to submit the manuscript for publication. Dr. Pat
the Lion’s Medical Research Foundation, a Leo J. Mahar Electrophysiology
Australian Postgraduate Award from the University of Adelaide. Ms. Middel
from the University of Adelaide. Dr. Mahajan is supported by the Leo J. Ma
Twomey is supported by the Leo J. Mahar Electrophysiology Scholarship fro
by a Rhodes scholarship and a Postgraduate Medical Scholarship from the
Australia. Dr. Hendriks is supported by the Derek Frewin Lectureship fro
supported by the National Heart Foundation of Australia. Dr. Kalman i
National Health and Medical Research Council of Australia; served on th
Scientific; and received research funding from St Jude Medical, Biosense W
supported by a Postdoctoral Fellowship from the National Health and Med
Craig Lectureship from the University of Adelaide. Dr. Sanders is supported
by Practitioner Fellowships from the National Health and Medical Researc
board of Biosense Webster, Medtronic, CathRx, and St Jude Medical; rece
Webster, Medtronic, St Jude Medical, and Boston Scientific; and received
Boston Scientific, Biotronik, and Sorin.
All authors attest they are in compliance with human studies committe
institutions and Food and Drug Administration guidelines, including pa
mation, visit the JACC: Clinical Electrophysiology author instructions page
Manuscript received August 25, 2016; revised manuscript received Decembe
risk of complications, and direct medicalcosts (21–24). Aggressive management ofthese risk factors in a dedicated physician-ledclinic has been shown to reduce the burdenof AF and improve the long-term success ofablation (25–27).
In the LEGACY (Long-Term Effect of GoalDirected Weight Management in an AtrialFibrillation Cohort: A Long-Term Follow-UpStudy) study, progressive weight-loss had adose-dependent effect on long-term freedomfrom AF (28). However, it is not clear if adedicated risk factor management (RFM)
clinic is cost-effective. In this study, we aim to eval-uate the cost and clinical effectiveness of a dedicatedRFM clinic in overall management of AF.
METHODS
STUDY POPULATION. The impact of weight lossand its effects on AF outcomes from our registrywere presented in the LEGACY Study (28). In theLEGACY study, all suitable patients (with body massindex $27 kg/m2 and $1 risk factor) were offeredRFM in a dedicated physician-directed clinic at thetime of initial assessment (Figure 1). Here we comparethe clinical and cost-effectiveness of a dedicatedRFM clinic for long-term results of patients diagnosedwith AF. Patients were dichotomized based onwhether they accepted this strategy and formed theintervention group (RFM group), whereas those whodeclined formed the control group. The study proto-col was approved by the Human Research EthicsCommittee of the Royal Adelaide Hospital and Uni-versity of Adelaide, Adelaide, Australia.
hak is supported by a Postgraduate Scholarship from
Scholarship from the University of Adelaide, and an
dorp is supported by the Robert J. Craig Scholarship
har Lectureship from the University of Adelaide. Dr.
m the University of Adelaide. Dr. Wong is supported
National Health and Medical Research Council of
m the University of Adelaide. Dr. Abhayaratna is
s supported by Practitioner Fellowships from the
e advisory board of Biosense Webster and Boston
ebster, Medtronic, and Boston Scientific. Dr. Lau is
ical Research Council of Australia and by a Robert J.
by the National Heart Foundation of Australia and
h Council of Australia; has served on the advisory
Pathak et al. J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 3 , N O . 5 , 2 0 1 7
Cost-Effectiveness of Risk Factor Management in AF M A Y 2 0 1 7 : 4 3 6 – 4 7
438
RISK FACTOR MANAGEMENT. Patients in the RFMgroup attended a physician-directed RFM clinic(in addition to their arrhythmia follow-up) at leastevery 3 months and were encouraged to use supportcounselling and to schedule more frequent reviews asrequired. Risk factors were managed according toAmerican College of Cardiology/American HeartAssociation guidelines. The details of our RFM havebeen previously presented (28). In brief, a structuredmotivational, goal-directed program using face-to-face counseling was used to achieve behavioralchange for weight management and increa-sing physical activity. Weight, hypertension,glucose intolerance, dyslipidemia, sleep apnea,and alcohol and tobacco use were screened andmanaged individually according to American Collegeof Cardiology/American Heart Association guidelines.The control group was given information on manage-ment of risk factors and encouraged to begin RFMunder the direction of their treating physician.
ARRHYTHMIA MANAGEMENT. Management of AFwas undertaken in a separate arrhythmia clinic by
physicians blinded to the patients’ study group.The use of rate and rhythm control strategies was atthe discretion of the treating physician. In patientswho remained symptomatic despite the use of anti-arrhythmic agents, AF ablation was offered. Theablation technique used at our institution has beenpreviously described (29). Arrhythmia occurrence wasdetermined by clinical review, 12-lead electrocardio-gram, and 7-day Holter monitoring at yearly intervalsbefore ablation. AF was taken as any atrialarrhythmia $30 s. Following ablation, patients weresimilarly reviewed every 3 months for the first yearand every 6 months thereafter. If patients developedrecurrent symptomatic arrhythmia after the blankingperiod (3 months), repeat ablation was offered.All patients were anticoagulated for CHADS2 score >1.Amiodarone was not used routinely and if requiredonly for 3- to 6-month periods. No patient continuedon amiodarone after ablation.
OUTCOMES FOR CLINICAL EFFECTIVENESS. Theprimary outcome was AF symptom burden as deter-mined by the AF Severity Scale (AFSS, University of
FIGURE 2 Decision Tree for Risk Factor Management Group
Decision tree used to calculate probability of patients remaining AF at annual follow-up and undergoing AF ablation in risk factor management
group. Online Figure 1 provides actual numbers observed and the complete decision tree for both groups. AF ¼ atrial fibrillation.
J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 3 , N O . 5 , 2 0 1 7 Pathak et al.M A Y 2 0 1 7 : 4 3 6 – 4 7 Cost-Effectiveness of Risk Factor Management in AF
439
Toronto), which quantifies 3 domains of AF-relatedsymptoms: 1) frequency; 2) duration; and 3)severity. The AFSS has been clinically validated andused for assessment of AF burden (30). The AFSSquestionnaire was administered at baseline andfinal follow-up. Freedom from AF was ascertainedwith 7-day Holter monitoring. Secondary outcomesincluded healthcare use (hospitalization, emergencydepartment presentation, unscheduled specialistclinic presentation), medication use (antihypertensivemedication, lipid therapy, antiarrhythmic use),sleep apnea device requirement, and procedural re-quirements (cardioversion, echocardiography, trans-esophageal echocardiography, ablation procedure).
OUTCOMES FOR COST EFFECTIVENESS. Quality-adjustedlife years. At each annual follow-up, freedom from AFwas recorded and converted into a utility value inaccordance with previously validated and publisheddata (7,31,32). The utility value for “Not AF Free” was0.725. Using an increase in utility value of 0.065 forthe successful treatment for AF, we generated a
utility value of 0.79 for the health state “AF Free.”A 3% discount rate was applied to the annual utilityvalues. Details of the utility value calculation and thediscount rate used for the study is provided in theOnline Appendix.
Model structure and modeling framework. Short-termanalysis was undertaken using a decision tree(Figure 2). Study data provided the probabilities ofpatients undergoing ablation procedures and thelikelihood that, for each pathway, the patient wouldachieve “AF Freedom” at Year 4. Incremental cost-effectiveness ratios (ICER) were calculated for costper quality-adjusted life years (QALY) gained(Figure 3). Sensitivity analyses were performed andplotted in a tornado diagram (Figure 4). Bootstrappingof 1,000 pair-wise comparisons improved estimates ofthe sampling distribution and provided data for thecost-effectiveness utility plane.
A Markov model estimated the long-term cost-effectiveness of RFM over 10 years, correspondingto the longest follow-up of ablation outcomes
Incremental cost-effectiveness ratios graph for the bootstrapping of the 1,000 pair-wise
comparison. Risk factor management intervention is in the South-Eastern quadrant, which
demonstrates that it is not only more effective than the comparison/control strategy, but
is also cost saving, and as a result, falls well below the cost-effectiveness threshold for a
new intervention. AUD ¼ Australian Dollar; QALY ¼ quality-adjusted life years.
Pathak et al. J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 3 , N O . 5 , 2 0 1 7
Cost-Effectiveness of Risk Factor Management in AF M A Y 2 0 1 7 : 4 3 6 – 4 7
440
available (17). The health states used were “AF Free”and “Not AF Free.” There were no deaths or strokes inpatients, and the health state “Dead” was notincluded. All patients entered the Markov model inthe health state “Not AF Free.” Using the Markovianassumption, annual transition probabilities werecalculated using an average of recorded results in thestudy for each cohort. In each annual cycle, patientscould remain “Not AF Free” (with transition proba-bilities of RFM, 48%; control, 72%), or transition to“AF Free” (RFM, 52%; control, 28%). Once “AF Free”patients could remain in this state (RFM, 75%; con-trol, 34%) or transition back to “Not AF Free” (RFM,25%; control, 66%).Costs . Using a bottom-up costing method, all costswere calculated in year 2010 Australian dollars.Costs were classified into 5 categories: 1) interven-tional procedures; 2) diagnostic procedures; 3) inpa-tient care; 4) outpatient visits; and 5) medication.Hospitalization costs were calculated using “AverageLength of Stay” for AF with the standard price perOccupied Bed Day, provided by the South AustralianDepartment of Health and Ageing. For each patient,the total number of drugs prescribed for arrhythmia,hypertension, and lipid disorder was recorded atbaseline and at follow-up. The price of specific brandsof medications, at standard dosages, was averaged
to provide an average annual cost for each drug(Online Tables 1 and 2).
For theMarkovmodel the annual health cost of eachhealth state was developed using the costs recorded inthe decision tree analysis. For each health state, theprobability of a patient requiring a clinic serviceannually was estimated, based on study data andstandard clinical knowledge. The annual probabilityof a clinical service was multiplied by the annual costto create a total annual cost per patient, and thensummed for each health state (Online Figure 2). Theannual cost for a patient in the state “AF Free” was$1,135 and “Not AF Free” was $5,207. Details areprovided in the Online Appendix (Online Tables 3 to 7).
STATISTICAL ANALYSIS. Categorical variables arerepresented by frequencies and percentages. Contin-uous variables are summarized by mean � SD. Thedifferences in baseline characteristics between groupswere assessed using analysis of variance proceduresfor continuous variables and chi-square test for cate-gorical variables. A repeated measure analysis ofvariance was used to assess change over time. Forcategorical variables, change in status at follow-upwas compared between groups using a chi-squaretest. Time-to-recurrence and event-free survivalcurves following the last ablation procedure wereestimated by the Kaplan-Meier product-limit method.Differences between curves were tested with the log-rank test. Predictors of recurrent AF were assessedusing proportional hazards Cox regression models.The proportional hazards assumption met for all themultivariable models that have been fit. The assump-tion was assessed using the Cox-Snell and Schoenfeldresiduals. Exploratory analysis of the univariate pre-dictors was carried out to understand the associationbetween recurrent AF and the covariates. MultiplesCox regression models were fit and assessed usingWald statistic and Akaike information criterion (AIC)to arrive at the final multivariate regression model.Two-tailed p < 0.05 was considered statistically sig-nificant. Statistical analysis was performed with SPSSversion 21.0 (SPSS, Inc., Chicago, Illinois).
RESULTS
BASELINE CHARACTERISTICS. Of the 1,415 consecutivepatients with symptomatic AF, 825 patients had bodymass index $27 kg/m2. After screening for exclusioncriteria, the final cohort comprised 355 patients; 208RFM and 147 control subjects (Figure 1). Mean follow-up in the RFM and control groups were 47.03 �17.9 months and 49.01 � 17.6 months, respectively
Tornado diagram sensitivity analysis showing the impact of change in each variable on the ICER per QALY. ICER ¼ incremental cost-effectiveness ratio;
other abbreviations as in Figures 1 and 3.
J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 3 , N O . 5 , 2 0 1 7 Pathak et al.M A Y 2 0 1 7 : 4 3 6 – 4 7 Cost-Effectiveness of Risk Factor Management in AF
441
(p ¼ 0.30). Baseline characteristics were similar in the2 groups (Table 1).
RISK FACTOR MODIFICATION. Table 2 shows theimpact of RFM on various cardiac risk factors. Therewas a greater decline in systolic blood pressure withRFM compared with control subjects (�10.1 �1.2 mm Hg vs. �3.3 � 1 mm Hg; p < 0.001). Weight andbody mass index decreased in both groups, butsignificantly more with RFM compared with controlsubjects (�10.1 � 8.8 kg vs. �3.3 � 8.4 kg; p < 0.0001)(Table 2). At baseline, 49% of RFM and 44% of controlsubjects had dyslipidemia (p ¼ 0.35). With diet andlifestyle modification, low-density lipoproteincholesterol and non-high-density lipoprotein choles-terol were well controlled in 47% of RFM and 15% ofcontrol subjects (p ¼ 0.02). At baseline, 30% of RFMand 27% of control subjects had a history of diabetesmellitus (p ¼ 0.36). At final follow-up, patients withdiabetes mellitus in the RFM group had significantlybetter glycemic control compared with control sub-jects (hemoglobin A1c <7% in 71% vs. 52%, respec-tively; p ¼ 0.003). At baseline, 53% of RFM and 48% ofcontrol subjects had severe obstructive sleep apnea(apnea-hypopnea index $30; p ¼ 0.35). At final followup, 31% in the RFM group and 39% in the control grouphad severe obstructive sleep apnea (p ¼ 0.003) sug-gesting significant improvement in the RFM group.
EFFECT OF RFM CLINIC ON AF SYMPTOM
BURDEN. At baseline, both groups had comparable
and high AFSS subscale scores (Table 2). AF fre-quency, duration, symptom, and symptom severitywere reduced at final follow-up in both groups with asignificantly greater reduction seen in the RFM group(p < 0.001).Freedom from AF without the use of rhythmcontro l st ra teg ies . Figure 5A demonstrates the“ablation and drug free” AF freedom based on groups.At final follow-up, 35% of patients in the RFM groupremained free from arrhythmia without antiar-rhythmic drugs or ablation, compared with 18% ofpatients in the control group (p < 0.001). Univariatepredictors of AF recurrence without antiarrhythmicdrugs or ablation were as follows: control group(hazard ratio [HR]: 1.6; 95% confidence interval [CI]:1.2 to 2.0; p < 0.001) and diastolic dysfunction (HR:1.5; 95% CI: 1.2 to 1.7; p < 0.001). On multivariateanalysis: control group (HR: 1.7; 95% CI: 1.3 to 1.8;p < 0.001) and diastolic dysfunction (HR: 1.4; 95% CI:1.2 to 1.8; p < 0.001) were independently associatedwith increased risk of AF recurrence.Total arrhythmia-free survival. Figure 5B demonstratesthe arrhythmia-free survival after multiple proced-ures with a significant attrition in control subjectscompared with RFM. At final follow-up, thearrhythmia-free survival rate following the lastcatheter ablation procedure was 79% with RFMcompared with 44% for control subjects (p < 0.001).Univariate predictors of AF recurrence aftermultiple procedures were as follows: control group
AF ¼ atrial fibrillation; AHI ¼ apnea-hypopnea index; BMI ¼ body mass index; DM ¼ diabetesmellitus; IGT ¼ impaired glucose tolerance; LA ¼ left atrium; LV IVS ¼ left ventricular interven-tricular septum; LVEDD ¼ left ventricular end diastolic diameter; RFM ¼ risk factor management.
Pathak et al. J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 3 , N O . 5 , 2 0 1 7
Cost-Effectiveness of Risk Factor Management in AF M A Y 2 0 1 7 : 4 3 6 – 4 7
442
(HR: 3.3; 95% CI: 2.3 to 4.9; p < 0.001); diabetes(HR: 1.8; 95% CI: 1.2 to 2.6; p ¼ 0.002), and tobacco(HR: 2.1; 95% CI: 1.12 to 4.25; p ¼ 0.034). Controlgroup (HR: 3.6; 95% CI: 2.4 to 5.2; p < 0.001)remained the most significant predictor of recurrentAF in multivariate analyses.
EFFECT OF RFM CLINIC ON HEALTH CARE USE. Table 3shows the details of the healthcare use betweengroups over the follow-up period.Ablat ion . Of the patients who required AF ablation,the mean number of AF ablation procedures per-formed in the RFM group was 0.60 � 0.69 and incontrol subjects was 0.72 � 0.86 (p ¼ 0.09). Of thepatients who were AF free at final follow up, 35% ofRFM patients had no procedure versus 18% in thecontrol group (p < 0.001), 41% of RFM patientsrequired only a single procedure versus 29% of con-trol subjects (p ¼ 0.02), and 9% of RFM patientsrequired multiple procedures versus 29% of controlsubjects (p ¼ 0.005).Cardioversion. The average number of cardioversionsper patient was 0.89 � 1.5 in the RFM group and
1.51 � 2.3 in the control group (p ¼ 0.002). Forty-eight (23%) patients in the RFM group had >1cardioversion versus 54 (37%) in the control group(p ¼ 0.02).
Medicat ions . Mean antihypertensive medicationuse in the RFM group was 0.53 � 0.7 and in the con-trol group 0.78 � 0.6 (p ¼ 0.04). At final follow-up,mean antiarrhythmic use in the RFM group was0.26 � 0.5 and in the control group was 0.91 � 0.6(p ¼ 0.003). At final follow up, 11% of the RFMpatients versus 35% in control group were on antilipidtherapy at final follow-up (p ¼ 0.008).
Emergency presentat ions . The average number ofemergency department presentations caused by anepisode of AF was 0.18 � 0.5 in the RFM group and0.76 � 1.2 among the control subjects (p < 0.001). Inthe RFM group 30 (14%) had 1 presentation and 6 (3%)had 2 or more presentations, whereas 26 (18%) in thecontrol group had 1 and 29 (20%) had 2 or morepresentations.
Hosp i ta l admiss ions . The mean number of hospitaladmissions caused by AF was 0.74 � 1.3 in the RFMgroup and 1.05 � 1.6 (p ¼ 0.03) in the control group.
Unscheduled spec ia l i s ts v i s i t . The mean numberof unscheduled visits (caused by arrhythmia andassociated symptoms) to specialist clinics in the RFMgroup was 0.19 � 0.4 and in the control group was1.94 � 2.0 (p < 0.001).
COST-EFFECTIVENESS ANALYSIS. Qual i ty of l i fe . Inthe RFM group the QALY gain over 4 years was 2.74 �0.07 and in the control group this was 2.68 � 0.05(p ¼ 0.001). The decrease in AF burden in the RFMgroup was 9.54 � 5.49 and in control subjects this was6.31 � 4.79 (p < 0.001).
Costs . The estimates of health care costs are pre-sented in the Online Table 3. The mean expectedhealth care cost per patient over the 4-year study waslower in the RFM group ($17,421 � $9,073), comparedwith the control group ($20,388 � $7,870).
Cost-effect iveness . Based on the decision treemodel, the RFM group had a QALY gain of 0.06(p ¼ 0.001) and a cost saving of $2,968 � $438 perpatient, in comparison with the control group. Thisequates to an ICER of $52,305 saved per QALY gained.Using bootstrap replication, RFM had an increase of0.0568 QALYs and a cost saving of $3,012 � $1,599($53,452 saved per QALY gained). RFM was a domi-nant intervention, with ICERs predominantly in theSouth-Eastern quadrant of the cost-effective plane(Figure 3). Projecting over a 10-year model, RFMwould have an increase of 0.1930 QALYs and a costsaving of $12,094 (ICER of $62,653 saved per QALYgained).
Values are mean � SD or n (%). Impact of risk factor management through a dedicated clinic on cardiac risk factors, cardiac structure, and AF severity from baseline to follow-up. *Mean follow-up: 47.03 �17.9 for RFM group and 49.01 � 17.6 months for control group. †p value refers to within-group difference (baseline to follow-up). ‡p value refers to between-group differences over time (group-timeinteraction).
MET ¼ metabolic equivalent; SBP ¼ systolic blood pressure; other abbreviations as in Table 1.
J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 3 , N O . 5 , 2 0 1 7 Pathak et al.M A Y 2 0 1 7 : 4 3 6 – 4 7 Cost-Effectiveness of Risk Factor Management in AF
443
SENSITIVITY ANALYSES. Sensitivity analyses wereundertaken on factors that could potentially affectthe cost-effectiveness of RFM, including proceduralcosts, the number and cost of clinics attended, thenumber of hospital admissions for AF-related causes,and the probability of undergoing initial and repeatprocedures (Figure 4). Cost-effectiveness was mostsusceptible to ablation procedure cost, with anincrease in price resulting in higher savings per QALYgained; and the probability of a patient requiring AFablation, with a higher probability resulting in lowersavings.
DISCUSSION
This study demonstrates that in overweight andobese individuals with highly symptomatic AF, astructured physician-directed risk factor and weightmanagement program was not only clinically effec-tive, improving AF outcomes, quality of life mea-sures, and health care use, but was also cost saving.These findings underscore the clinical importanceand significant cost-effectiveness of treating the un-derlying causes of AF to achieve rhythm control andmaintenance of sinus rhythm.
This study was an attempt to determine the cost-effectiveness of aggressive RFM as a concurrenttreatment strategy in patients with AF. Our resultsdid not compare the clinical and cost-effectiveness of
RFM and AF ablation as treatment options exclusiveof each other. Moreover, we did not focus on cost-effectiveness regarding rate versus rhythm control.However, this study does suggest the order in whichtreatment should be offered. In the study, successfulRFM was associated with a 38% reduction in need forinitial AF ablation procedure and a 20% reduction inneed for redo ablation. In sensitivity analyses, costswere most sensitive to the need for initial and redoAF ablation procedures. RFM strategy is associatedwith higher upfront cost because of increased clinicvisits. In the long term, reducing the number ofpatients requiring ablation procedures will reduceoverall costs substantially. However, if ablationcannot be avoided, RFM should still be offered toreduce the need for redo procedures by improving theablation outcomes. Thus, our results show that RFMshould be offered as first-line therapy to patients withAF. This study also indicates that RFM in patientswith AF is a cost-saving measure irrespective of theAF management strategy adopted.
Hospitalizations represent the major cost driver inAF care. Recent studies have found a 23% to 125%growth in AF hospitalization (8,9). The greater inci-dence of AF with increasing age is an obviouscontributor. However, the age-specific rate of AFhospitalizations is also increasing, possibly becauseof a worsening risk factor profile (33,34) and the costof AF hospitalizations increases proportionately with
FIGURE 5 AF Freedom Outcome According to Group
(A) Kaplan-Meier curve for AF-free survival without the use of rhythm-control strategies. (B) Kaplan-Meier curve for AF-free survival for total
AF-free survival (multiple ablation procedures � drugs; right). Abbreviations as in Figure 1.
TABLE 3
Medicatio
Antiarrh
Antihyp
Lipid th
CPAP us
Interventi
Cardiov
Single A
Second
Third pr
In-patient
ED pres
Hospita
Out-patien
Speciali
Speciali
Speciali
Values are mwithin-grou
CPAP ¼abbreviation
Pathak et al. J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 3 , N O . 5 , 2 0 1 7
Cost-Effectiveness of Risk Factor Management in AF M A Y 2 0 1 7 : 4 3 6 – 4 7
444
an increase in CHADS2 score (9). In this study, wefound that RFM was associated with a 36% reductionin hospitalization. In addition, even if managed withrate control strategies the cost of AF care is still
Health Care Use
Control Group(N¼147)
RFM Group(N¼208) p Value
n use
ythmic use 0.91 � 0.6 0.26 � 0.5 0.003
ertensive medication 0.78 � 0.6 0.53 � 0.7 0.043
erapy 81 (64) 73 (40) 0.032
e 62 (43) 70 (34) 0.098
onal requirement
ersion 1.51 � 2.3 0.89 � 1.5 0.002
F ablation 43 (29) 86 (41) 0.009
procedure 24 (16) 15 (7)
ocedure 5 (3) 3 (1)
visits
entation 0.76 � 1.2 0.18 � 0.5 <0.001
lization for AF 1.15 � 1.6 0.74 � 1.3 0.034
t visits
st, planned 14 � 3 10 � 2 0.01
st visit, RFM clinic Nil 16 � 4 <0.001
st visit, unplanned 1.94 � 2.0 0.19 � 0.4 <0.001
ean � SD or n (%). Mean health care use in control and RFM groups. P value refers top difference (baseline to follow-up).
continuous positive pressure ventilation; ED ¼ emergency department; others as in Table 1.
driven by medication use, emergency departmentpresentation, and specialist appointments. We foundRFM improved the overall health care use with a 20%reduction in medication use, a 58% reduction inemergency department presentation, and a 1.8-foldreduction in unscheduled specialist reviews.
Catheter ablation is an effective therapy for rhythmcontrol in patients with drug-refractory or intolerantAF and is increasingly used (35). Although it is asso-ciated with upfront costs it is more effective inmaintaining sinus rhythm with a downstream costsaving, because of avoided hospitalizations, reducedantiarrhythmic drugs use, and greater improvementin quality of life (36,37). However, clinical and cost-effectiveness studies of catheter ablation of AF aresensitive to the time horizon for the analysis (38).Short time horizons bias against ablation because ofthe high initial costs associated with the procedure.In contrast, there is gradual attrition of success afterablation with time, reducing the clinical effectivenessand also the cost-effectiveness (7). From a long-termperspective, the cost-effectiveness of catheter abla-tion of AF depends on single-procedure efficacy.Cardiac risk factors are associated with increased riskof AF recurrence post-ablation, increasing need forredo procedures (21,22,39–41). RFM improves thelong-term freedom from AF and thus improves thecost and clinical effectiveness of the AF ablation.
PERSPECTIVES
COMPETENCY IN MEDICAL KNOWLEDGE: Risk factor and
weight management are essential components of the manage-
ment of the patient with AF. This approach is clinically effective
and cost saving. A dedicated physician-led clinic that is focused
on weight and risk factor management is important to achieve
long-term success.
COMPETENCY IN PATIENT CARE: Risk factor and weight
management are crucial to a strategy of rhythm control in pa-
tients with AF. A dedicated clinic improves patient engagement
by promoting treatment adherence and improves long-term
outcomes.
TRANSLATIONAL OUTLOOK: Risk factor and weight man-
agement are essential components of the management of the
patient with AF. Considering the dual epidemic of obesity and AF,
primary and secondary prevention strategies should be
increasingly used. A dedicated weight and RFM clinic may play a
pivotal role in this regard.
J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 3 , N O . 5 , 2 0 1 7 Pathak et al.M A Y 2 0 1 7 : 4 3 6 – 4 7 Cost-Effectiveness of Risk Factor Management in AF
445
We used quality-of-life adjustment as it pertains tosinus rhythm maintenance, which permits theexpression of results in dollars per QALY. RFM wasassociated with a gain in QALYs at a cost saving of$53,452.
These results were insensitive to changes in the costestimates explored in the sensitivity analyses. Theresults of the sensitivity analysis in some way enableus to compare the cost-effectiveness of the RFM indifferent health systems. The cost-effectiveness ofRFM was evaluated for the Australian health system,where average cost of AF ablation is calculated at$13,847. Sensitivity analysis was undertaken, with thecost of the procedure taken from $6,000 to $20,000.RFM became incrementally cost-effective with highercost of AF ablation. In the United States, there is a widesite level variation in cost of AF ablation. Althoughthere is paucity of published data in this regards, it isestimated that it ranges from $16,200 to $$77,300(42,43). Resultantly, RFM may be even more cost-effective in the U.S. health care system.
In this study, aggressive RFM was more clinicallyeffective and less costly than usual care. However, thetotal costs of AF care based on the previouslymentioned factors are probably an underestimation ofthe true costs of AF. The indirect societal costs relatedto lost productivity and the time costs associatedwith primary care and specialist visits are consider-able. Although not assessed in the current study,if the reduced AF burden seen in this study overthe long term is associated with reduced risk ofdementia, stroke, and institutionalization of elderly,the cost-effectiveness of RFM may be even morefavorable.
STUDY LIMITATIONS. This study has the usual po-tential for selection bias inherent in observationalstudies. However, measurement bias has beenreduced through standardized processes in our clinicand the evaluation by operators blinded to thepatient’s weight management strategy. AF burdenassessment using 7-day Holter may not detect some AFepisodes, especially in patients with low AF burden.However, this was used in both groups and thusunlikely to introduce detection bias. Ascertainmentbias was reduced through the collection of outcome viaroutine data sources. In addition, our population wasrelatively young, therefore these findings may not beapplicable for the elderly population. As is invariablythe case with cost-effectiveness modeling studies,simplifying assumptions and a degree of parameteruncertainty exists. We also acknowledge that thestudies used for utility value calculation are not acomplete match for our study cohort and intervention.
However, we think the utility value for the “AF not freestate” and “AF-free state” can still be used becauseboth these health states remain relevant irrespectiveof the type of intervention used to achieve them.Moreover, with the type of intervention (RFM) leadingto AF-free state in this study, the gain in utility valuemay even be a conservative estimate when comparedwith AF-free state achieved with drug therapy orablation. A further limitation of this cost analysis isthat calculations are based on economic conditions in asingle country. Given the absence of such events in ourcohort, no assessment was made on the risk of strokeor death and their associated costs.
CONCLUSIONS
A structured physician-directed risk factor andweight management program is effective and resultsin the long-term freedom from AF. This approach isclinically effective and cost saving. With the growingepidemic of AF and the overall health care costburden, this strategy should be increasingly used.
ADDRESS FOR CORRESPONDENCE: Dr. PrashanthanSanders, Centre for Heart Rhythm Disorders, Depart-ment of Cardiology, Royal Adelaide Hospital, L5McEwin Building, North Terrace, Adelaide, SouthAustralia 5000, Australia. E-mail: [email protected].
Pathak et al. J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 3 , N O . 5 , 2 0 1 7
Cost-Effectiveness of Risk Factor Management in AF M A Y 2 0 1 7 : 4 3 6 – 4 7
446
RE F E RENCE S
1. Chugh SS, Havmoeller R, Narayanan K, et al.Worldwide epidemiology of atrial fibrillation: aGlobal Burden of Disease 2010 Study. Circulation2014;129:837–47.
2. Wattigney WA, Mensah GA, Croft JB. Increasingtrends in hospitalization for atrial fibrillation in theUnited States, 1985 through 1999: implicationsfor primary prevention. Circulation 2003;108:711–6.
3. Miyasaka Y, Barnes ME, Bailey KR, et al. Mor-tality trends in patients diagnosed with first atrialfibrillation: a 21-year community-based study.J Am Coll Cardiol 2007;49:986–92.
4. Kannel WB, Wolf PA, Benjamin EJ, Levy D.Prevalence, incidence, prognosis, and predispos-ing conditions for atrial fibrillation: population-based estimates. Am J Cardiol 1998;82:2N–9N.
5. Kim MH, Johnston SS, Chu BC, Dalal MR,Schulman KL. Estimation of total incrementalhealth care costs in patients with atrial fibrillationin the United States. Circ Cardiovasc Qual Out-comes 2011;4:313–20.
6. Le Heuzey JY, Paziaud O, Piot O, et al. Cost ofcare distribution in atrial fibrillation patients: theCOCAF study. Am Heart J 2004;147:121–6.
7. Reynolds MR, Essebag V, Zimetbaum P,Cohen DJ. Healthcare resource utilization andcosts associated with recurrent episodes of atrialfibrillation: the FRACTAL registry. J CardiovascElectrophysiol 2007;18:628–33.
8. Wong CX, Brooks AG, Leong DP, Roberts-Thomson KC, Sanders P. The increasing burden ofatrial fibrillation compared with heart failure andmyocardial infarction: a 15-year study of all hos-pitalizations in Australia. Arch Intern Med 2012;172:739–41.
9. Patel NJ, Deshmukh A, Pant S, et al. Contem-porary trends of hospitalization for atrial fibrilla-tion in the United States, 2000 through 2010:implications for healthcare planning. Circulation2014;129:2371–9.
10. Tedrow UB, Conen D, Ridker PM, et al. Thelong- and short-term impact of elevated bodymass index on the risk of new atrial fibrillation theWHS (women’s health study). J Am Coll Cardiol2010;55:2319–27.
11. Gami AS, Hodge DO, Herges RM, et al.Obstructive sleep apnea, obesity, and the risk ofincident atrial fibrillation. J Am Coll Cardiol 2007;49:565–71.
12. Singh SN, Tang XC, Singh BN, et al. Quality oflife and exercise performance in patients in sinusrhythm versus persistent atrial fibrillation: a Vet-erans Affairs Cooperative Studies Program Sub-study. J Am Coll Cardiol 2006;48:721–30.
13. Corley SD, Epstein AE, DiMarco JP, et al. Re-lationships between sinus rhythm, treatment, andsurvival in the Atrial Fibrillation Follow-Up Inves-tigation of Rhythm Management (AFFIRM) Study.Circulation 2004;109:1509–13.
14. Hagens VE, Ranchor AV, Van Sonderen E, et al.Effect of rate or rhythm control on quality of life inpersistent atrial fibrillation. Results from the RateControl Versus Electrical Cardioversion (RACE)Study. J Am Coll Cardiol 2004;43:241–7.
15. Calkins H, Kuck KH, Cappato R, et al. 2012HRS/EHRA/ECAS expert consensus statement oncatheter and surgical ablation of atrial fibrillation:recommendations for patient selection, proceduraltechniques, patient management and follow-up,definitions, endpoints, and research trial design:a report of the Heart Rhythm Society (HRS) TaskForce on Catheter and Surgical Ablation of AtrialFibrillation. Heart Rhythm 2012;9:632–96.
16. Weerasooriya R, Khairy P, Litalien J, et al.Catheter ablation for atrial fibrillation: are resultsmaintained at 5 years of follow-up? J Am CollCardiol 2011;57:160–6.
17. Steinberg JS, Palekar R, Sichrovsky T, et al.Very long-term outcome after initially successfulcatheter ablation of atrial fibrillation. HeartRhythm 2014;11:771–6.
18. Ganesan AN, Shipp NJ, Brooks AG, et al. Long-term outcomes of catheter ablation of atrialfibrillation: a systematic review and meta-analysis.J Am Heart Assoc 2013;2:e004549.
19. Shah RU, Freeman JV, Shilane D, Wang PJ,Go AS, Hlatky MA. Procedural complications,rehospitalizations, and repeat procedures aftercatheter ablation for atrial fibrillation. J Am CollCardiol 2012;59:143–9.
20. Khaykin Y, Wang X, Natale A, et al. Costcomparison of ablation versus antiarrhythmicdrugs as first-line therapy for atrial fibrillation: aneconomic evaluation of the RAAFT pilot study.J Cardiovasc Electrophysiol 2009;20:7–12.
21. Shah AN, Mittal S, Sichrovsky TC, et al. Long-term outcome following successful pulmonaryvein isolation: pattern and prediction of very laterecurrence. J Cardiovasc Electrophysiol 2008;19:661–7.
22. Mohanty S, Mohanty P, Di Biase L, et al. Impactof metabolic syndrome on procedural outcomes inpatients with atrial fibrillation undergoing cath-eter ablation. J Am Coll Cardiol 2012;59:1295–301.
23. Hoyt H, Bhonsale A, Chilukuri K, et al. Com-plications arising from catheter ablation of atrialfibrillation: temporal trends and predictors. HeartRhythm 2011;8:1869–74.
24. Aldhoon B, Wichterle D, Peichl P, Cihak R,Kautzner J. Complications of catheter ablation foratrial fibrillation in a high-volume centre with theuse of intracardiac echocardiography. Europace2013;15:24–32.
25. Pathak RK, Middeldorp ME, Lau DH, et al.Aggressive risk factor reduction study for atrialfibrillation and implications for the outcome ofablation: the ARREST-AF cohort study. J Am CollCardiol 2014;64:2222–31.
26. Pathak RK, Elliott A, Middeldorp ME, et al.Impact of CARDIOrespiratory FITness on
Arrhythmia Recurrence in Obese Individuals WithAtrial Fibrillation: the CARDIO-FIT Study. J AmColl Cardiol 2015;66:985–96.
27. Abed HS, Wittert GA, Leong DP, et al. Effect ofweight reduction and cardiometabolic risk factormanagement on symptom burden and severity inpatients with atrial fibrillation: a randomizedclinical trial. JAMA 2013;310:2050–60.
28. Pathak RK, Middeldorp ME, Meredith M, et al.Long-term effect of goal-directed weight man-agement in an atrial fibrillation cohort: a long-term follow-up study (LEGACY). J Am CollCardiol 2015;65:2159–69.
29. Stiles MK, John B, Wong CX, et al. Paroxysmallone atrial fibrillation is associated with anabnormal atrial substrate. J Am Coll Cardiol 2009;53:1182–91.
30. Dorian P, Jung W, Newman D, et al. Theimpairment of health-related quality of life inpatients with intermittent atrial fibrillation: im-plications for the assessment of investigationaltherapy. J Am Coll Cardiol 2000;36:1303–9.
31. Reynolds MR, Zimetbaum P, Josephson ME,Ellis E, Danilov T, Cohen DJ. Cost-effectiveness ofradiofrequency catheter ablation compared withantiarrhythmic drug therapy for paroxysmal atrialfibrillation. Circ Arrhythm Electrophysiol 2009;2:362–9.
32. Reynolds MR, Lavelle T, Essebag V, Cohen DJ,Zimetbaum P. Influence of age, sex, and atrialfibrillation recurrence on quality of life outcomesin a population of patients with new-onset atrialfibrillation: the Fibrillation Registry AssessingCosts, Therapies, Adverse events and Lifestyle(FRACTAL) study. Am Heart J 2006;152:1097–103.
33. Wong CX, Brooks AG, Lau DH, et al. Factorsassociated with the epidemic of hospitalizationsdue to atrial fibrillation. Am J Cardiol 2012;110:1496–9.
34. Steinberg BA, Holmes DN, Ezekowitz MD,et al. Rate versus rhythm control for managementof atrial fibrillation in clinical practice: results fromthe Outcomes Registry for Better InformedTreatment of Atrial Fibrillation (ORBIT-AF) regis-try. Am Heart J 2013;165:622–9.
35. Kumar S, Walters TE, Halloran K, et al. Ten-year trends in the use of catheter ablation fortreatment of atrial fibrillation vs. the use of cor-onary intervention for the treatment of ischaemicheart disease in Australia. Europace 2013;15:1702–9.
36. Ladapo JA, David G, Gunnarsson CL, et al.Healthcare utilization and expenditures in patientswith atrial fibrillation treated with catheter abla-tion. J Cardiovasc Electrophysiol 2012;23:1–8.
37. Reynolds MR, Walczak J, White SA, Cohen DJ,Wilber DJ. Improvements in symptoms and qualityof life in patients with paroxysmal atrial fibrillationtreated with radiofrequency catheter ablationversus antiarrhythmic drugs. Circ Cardiovasc QualOutcomes 2010;3:615–23.
J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 3 , N O . 5 , 2 0 1 7 Pathak et al.M A Y 2 0 1 7 : 4 3 6 – 4 7 Cost-Effectiveness of Risk Factor Management in AF
447
38. McKenna C, Palmer S, Rodgers M, et al. Cost-effectiveness of radiofrequency catheter ablationfor the treatment of atrial fibrillation in the UnitedKingdom. Heart 2009;95:542–9.
39. Jongnarangsin K, Chugh A, Good E, et al. Bodymass index, obstructive sleep apnea, and out-comes of catheter ablation of atrial fibrillation.J Cardiovasc Electrophysiol 2008;19:668–72.
40. Chao TF, Ambrose K, Tsao HM, et al. Rela-tionship between the CHADS(2) score and risk ofvery late recurrences after catheter ablation of
41. Pathak RK, Mahajan R, Lau DH, Sanders P. Theimplications of obesity for cardiac arrhythmiamechanisms and management. Can J Cardiol 2015;31:203–10.
42. Perino A, Fan J, Schmitt S, et al. Cost variationand associated outcomes of catheter ablation foratrial fibrillation. J Am Coll Cardiol 2015;65:A277.
43. Kim MH, Lin J, Kreilick C, Foltz Boklage SH.Total costs and atrial fibrillation ablation success
or failure in Medicare-aged patients in the UnitedStates. Adv Ther 2010;27:600–12.
KEY WORDS ablation, atrial fibrillation,cost effectiveness, risk factor management
APPENDIX For supplemental methods,tables, and figures, please see the onlineversion of this article.
