* Corresponding author. Tel.: ; fax: address: (IsabelleC. Van Gelder).

Transcription

1 European Heart Journal (2004) 25, Clinical research Rate control is more cost-effective than rhythm control for patients with persistent atrial fibrillation results from the RAte Control versus Electrical cardioversion (RACE) study Vincent E. Hagens a, Karin M. Vermeulen b, Elisabeth M. TenVergert b, Dirk J. Van Veldhuisen a, Hans A. Bosker c, Otto Kamp d, J. Herre Kingma e, Jan G.P. Tijssen f, Harry J.G.M. Crijns g, Isabelle C. Van Gelder a, *, for the RAte Control versus Electrical cardioversion for persistent atrial fibrillation study group a Department of Cardiology, Thoraxcenter, University Hospital Groningen, PO Box 30001, 9700 RB Groningen, The Netherlands b Office for Medical Technology Assessment, University Hospital, Groningen, The Netherlands c Department of Cardiology, Rijnstate Hospital, Arnhem, The Netherlands d Department of Cardiology, VU University Medical Center, Amsterdam, The Netherlands e Department of Cardiology, St. Antonius Hospital, Nieuwegein, The Netherlands f Department of Cardiology, Academical Medical Center, Amsterdam, The Netherlands g Department of Cardiology, University Hospital, Maastricht, The Netherlands Received 7 April 2004; revised 2 June 2004; accepted 10 June 2004 Available online 14 August 2004 KEYWORDS Atrial fibrillation; Costs; Rate control; Rhythm control Aims To evaluate costs between a rate and rhythm control strategy in persistent atrial fibrillation. Methods and results In a prospective substudy of RACE (Rate control versus electrical cardioversion for persistent atrial fibrillation) in 428 of the total 522 patients (206 rate control and 222 rhythm control), a cost-minimisation and cost-effectiveness analysis was performed to assess cost-effectiveness of the treatment strategies. After a mean follow-up of 2.3 ± 0.6 years, the primary endpoint (cardiovascular morbidity and mortality) occurred in 17.5% (36/202) of the rate control patients and in 21.2% (47/222) of the rhythm control patients. Mean costs per patient under rate control were 7386 and 8284 under rhythm control. Cost-effectiveness analysis showed that per avoided endpoint under rate control, the cost savings were Under rhythm control, more costs were generated due to electrical cardioversions, hospital admissions and anti-arrhythmic medication. Costs were higher in older patients, patients with underlying heart disease, those who reached a primary endpoint and women. Heart rhythm at the end of study, did not influence costs. * Corresponding author. Tel.: ; fax: address: i.c.van.gelder@thorax.azg.nl (IsabelleC. Van Gelder) X/$ - see front matter c 2004 The European Society of Cardiology. Published by Elsevier Ltd. All rights reserved. doi: /j.ehj

2 Rate control is more cost-effective than rhythm control 1543 Conclusions Rate control is more cost-effective than rhythm control for treatment of persistent atrial fibrillation. Underlying heart disease but not heart rhythm largely accounts for costs. c 2004 The European Society of Cardiology. Published by Elsevier Ltd. All rights reserved. Introduction Atrial fibrillation (AF) is the most common sustained cardiac arrhythmia. 1 It is estimated that its prevalence is likely to increase 2.5-fold during the next 50 years. 2 With this increasing number of subjects at risk for developing AF and its related morbidity like thrombo-embolic complications and heart failure, the socio-economic consequences may be overwhelming. 3 The number of hospital admissions for AF has already grown in recent years with increasing health care expenditures as a consequence. 4 6 Moreover, these hospitalisations are the main cost drivers in the treatment of AF. 7 With restricted health care resources and a growing number of patients, it is reasonable to evaluate costs and effects of different therapeutic options. Although rhythm control (i.e., restoration and maintenance of sinus rhythm) is often therapy of choice, rate control is eventually adopted since rhythm control is ineffective in the majority of patients. 8 A rhythm control approach is likely to enhance costs of AF since repeated cardioversions imply hospital admissions. On the other hand, rate control may possibly augment the occurrence of (or deterioration of) heart failure with increasing hospital admissions. Also, increase of palpitations and other AF related symptoms might enhance hospitalisations. Depending on differences in morbidity, mortality and quality of life between groups, costs may affect the decision whether to adopt a rate or rhythm control strategy. With comparable outcome between the treatment alternatives, the strategy generating the lowest costs is the treatment of choice. Four recent studies showed that mortality and morbidity were comparable between the two treatment strategies Furthermore, patients quality of life under both treatment strategies did not differ significantly. 9,13,14 Regarding cost-effectiveness, it remains unclear which treatment strategy is most cost-effective. As part of the Dutch RAte Control versus Electrical cardioversion (RACE) study a prospective evaluation of costs under rate and rhythm control was performed. 10 Methods Design of the RACE study The RACE study was a randomised study comparing long term effects of rate and rhythm control on morbidity and mortality in patients with persistent AF. The methods and primary outcome of this study have been recently published elsewhere. 10 In short, patients with persistent AF or atrial flutter, defined as non-self-terminating arrhythmia and requiring electrical cardioversion were included. 15 Under rate control, patients were treated with digitalis, a non-dihydropyridine calcium channel blocker, b-blocker alone or a combination of these. In the rhythm control group, patients underwent serial electrical cardioversion and serial anti-arrhythmic drug treatment using as first choice sotalol, thereafter class IC anti-arrhythmics, and at last amiodarone. From 4 weeks before until 4 weeks after the electrical cardioversion, all patients received acenocoumarol or fenprocoumon (International Normalized Ratio ). If sinus rhythm was present at 1 month, oral anti-coagulation could be stopped or changed to aspirin mg per day. Aspirin was also allowed in rate control patients below 65 years if they had AF without underlying cardiac disease. All other patients received oral anti-coagulation. The primary endpoint was the composite of cardiovascular mortality, heart failure, thrombo-embolic complications, bleeding, pacemaker implantation, or severe adverse effects of antiarrhythmic drugs. All patients at 31 centres in the Netherlands gave written informed consent. The Institutional Review Boards of all hospitals approved the study protocol. Study population In 428 patients (82% of the total of 522 patients), 206 rate and 222 rhythm control patients, an economic evaluation was performed which was a secondary endpoint of the study. Baseline characteristics are shown in Table 1. These were not significantly different from the originally included population of 522 patients. 10 There were no significant differences between both treatment groups except for a higher incidence of hypertension in the rhythm control group and of respiratory disease in the rate control group. 10 Economic evaluation and costs Mean follow-up of the study was 2.3±0.6 years. The RACE study showed that rate and rhythm control therapy in patients with persistent atrial fibrillation was equally effective regarding morbidity and mortality. 10 The primary endpoint occurred in 17.5% (36/206) of the patients randomised to rate control versus 21.2% (47/222) of the patients treated according to the rhythm control strategy (Table 2). This is comparable to the outcome of the original study that showed that rate control was not inferior to rhythm control: 17.8% (44/256) versus 22.6% (60/266). In RACE every primary endpoint was given a score prospectively to indicate its severity. The most severe endpoint (death) was given six points, an endpoint with no or incomplete recovery was given four points, and an endpoint with complete recovery was given one point. For both strategies we measured the total event burden by taking into account the number of endpoints and their severity. So, an event burden of 6 for the total group means that all patients died because of an endpoint. The event burden for the rate control strategy was 0.66 and 0.73 for rhythm control. This was not statistically different (p=0.38). Thus, also using this analysis, rate control was

3 1544 V.E. Hagens et al. Table 1 Patient characteristics at baseline Number (%), mean±sd, or median (range) Rate control (n=206) Rhythm control (n=222) P-value Age (years) 69±9 69± Men 129 (63) 139 (63) 0.99 Total AF duration (days) 479 (14 14,909) 410 (27 14,399) 0.54 Duration present episode of AF (days) 33 (1 399) 36 (1 395) 0.66 Underlying diseases Coronary artery disease 63 (31) 59 (27) 0.39 Old myocardial infarction 35 (17) 32 (14) 0.51 Valvular disease 39 (19) 35 (16) 0.43 Mitral valve disease 27 (13) 25 (11) 0.66 Aortic disease 12 (6) 11 (5) 0.83 Aortic and mitral valve disease 0 1 (1) 0.35 Cardiomyopathy 26 (13) 17 (8) 0.11 History of hypertension 83 (40) 127 (57) <0.01 History of respiratory disease 53 (26) 38 (17) 0.03 History of diabetes mellitus 26 (13) 22 (10) 0.44 No heart disease 41 (20) 40 (18) 0.63 NYHA class for heart failure 0.91 I 91 (44) 102 (46) II 108 (52) 112 (51) III 7 (3) 8 (4) History of Heart failure 115 (56) 120 (54) 0.89 Thrombo-embolic complication 35 (17) 27 (12) 0.17 Stroke 6 (3) 10 (5) 0.10 Haemorrhagic complications 16 (8) 20 (9) 0.73 AF, atrial fibrillation; NYHA, New York Heart Association functional class; SD, standard deviation. Table 2 Primary endpoint and its components by treatment group a Number (%) Rate control (n=206) Rhythm control (n=222) Primary endpoint 36 (17.5) 47 (21.2) Cardiovascular mortality 16 (7.3) 13 (5.9) Heart failure 8 (3.9) 9 (4.1) Thrombo-embolic complications 9 (4.4) 17 (7.7) Bleedings 10 (4.9) 5 (2.3) Adverse effects of anti-arrhythmic drugs 1 (0.5) 10 (4.5) Pacemaker implants 3 (1.5) 7 (3.2) a No double counts within lines, between lines double counts are possible. not inferior to rhythm control. In addition, patients quality of life, also in this substudy, did not differ significantly between both treatment arms during follow-up. 14 The present study was designed to test the explicit hypothesis of equivalence in outcome between rate and rhythm control. With comparable outcome regarding the occurrence of the primary endpoint and the quality of life in both treatment options for persistent atrial fibrillation, we performed a cost-minimisation analysis. 16,17 This means that the treatment option that generates the smallest amount of costs would be the most cost-effective and, consequently, the treatment of choice. The difference in costs between the two treatment strategies was calculated on a patient level. Furthermore, a cost-effectiveness ratio was calculated for the primary endpoint to explore the cost savings per avoided endpoint. Sensitivity analysis was performed to explore to what extent the total costs were influenced by varying costs on the different cost categories ( 20% versus +20%). The costs of therapy were compared during the following periods: between randomisation and 3 months, between 3 and 6 months, between 6 and 12 months, between 12 and 24 months, and between 24 and 30/36 months (end of study). In addition to the rate versus rhythm control comparison, costs were also evaluated in clinical relevant subgroups, irrespective of treatment strategy. The economic evaluation was performed from a societal perspective: all relevant direct costs both inside and outside the health care system, were taken into account during the total fol-

4 Rate control is more cost-effective than rhythm control 1545 low-up. Costs were divided into direct medical costs and direct non-medical costs. Data concerning costs of hospital stay, outpatient visits, cardioversions and medication were collected at the scheduled study visits in the outpatient department at 1, 3, 6, 12, 24 months after randomisation and at the end of the study (at months). These pre-set study visits were not taken into account for the cost analysis. Information on costs made outside the treatment centres, general practitioner visits, thrombosis laboratory, professional and non-professional help, and productivity losses were collected through self-administered patient questionnaires. Costs of a primary endpoint were only considered in the above mentioned cost categories. For instance, the costs of a pacemaker or a stent used in an acute coronary syndrome, were not recorded. However, costs for the hospital admission, post-intervention visits to the outpatient department, and other related costs were collected. For assessing the costs, Dutch standard prices were used (Table 3). 18 For costs of medication, which also included taxes (6%) and costs for the pharmacist ( 5.40 per prescription), the Pharmacotherapeutic Compass was used. 19 For electrical cardioversions, there were no standard prices available. Therefore, costs were estimated based on true resources used and time invested by the staff involved in the electrical cardioversion. Prices for these cardioversions were assessed at the University Hospital Groningen and at the Rijnstate Hospital, Arnhem (general hospital). These two centres were assumed to be representative for all participating university and general hospitals. The difference in kind of hospital was made, as costs of medical therapy were generally higher for a university hospital. Costs for electrical cardioversions were not considered for the item of hospital admissions unless the procedure made an overnight stay in the hospital necessary. The category of informal care means nonpaid help of the patient by family and friends. For this help, fixed prices for cleaning staff were used. The price level used was that of the year Prices have been assessed in Dutch guilders and were converted to Euros. According to the Dutch guidelines for cost-studies, costs made in the second and third year of the study were discounted to the start of the treatment. 18 The recommended discount rate of 4% was used in the present study. The economic evaluation was performed on an intention to treat basis, i.e., all costs were assigned to the treatment strategy to which the patients were first randomised. Table 3 Unit prices of included cost categories a Statistical analysis The Kolmogorov Smirnov test was used to compare the distribution of costs to the standard normal distribution. None of the cost-categories fitted the standard normal distribution. Therefore, differences between groups were tested with the Mann Whitney U test. A p-value <0.05 was considered statistically significant. Unfortunately, as is often the case in this type of analysis, some data were missing. Therefore, statistical testing was only possible for the separate time periods but not at the level of total resource use per patient. Furthermore, mean costs for each item of resource use per study period was calculated and than aggregated to estimate the total costs per patient. Results The mean total costs per patient for the whole study period for the rate control group were 7386 and 8284 for the rhythm control group (Table 4). This means that the rhythm control strategy generated extra costs to the amount of 898. A cost effectiveness ratio was estimated for the primary endpoint by calculating the difference in costs between rate and rhythm control divided by the difference in the occurrence of the primary endpoint. A discount rate of 4% was used for both cost and effects. Per avoided endpoint under rate control, the costs savings are ( / ). Costs during follow-up The first six months after randomisation generated a relatively large amount of costs. During these months, rhythm control was slightly more expensive due to (repeated) electrical cardioversion to restore sinus rhythm (Table 5). The costs for (a) cardioversion(s), with 303 procedures in 209 rhythm control patients versus 26 cardioversions in 21 rate control patients (b) medication, Cost category Unit price ( ) Method Cardioversion university hospital 258 Based on true costs: staff, equipment, medication, housing, housekeeping and overheads Cardioversion general hospital 153 Based on true costs: staff, equipment, medication, housing, housekeeping and overheads Medication Diverse Dutch Pharmacotherapeutic Compass Taxes (6%) and pharmacy costs (5.40 per prescription) included Outpatient visit university hospital Standard price outpatient visit university hospital Outpatient visit general hospital 42 Standard price outpatient visit general hospital Hospital admission university hospital 340 Standard price hospital day university hospital Hospital admission general hospital 242 Standard price hospital day general hospital General practitioner 2.30 Derived from standard price per minute Thrombosis laboratory 8.50 Derived from the College for Tariffs in Dutch Health Care District nurse b 0.38 Standard price district nursing per minute Paid help family b 23 Standard price paid help family Informal care 8 Standard price for cleaning staff Travel costs Diverse Standard price for car and public transport (based on the mean distances to the different health services) Productivity losses Diverse Friction costs; depends on income a All prices are from the price level of the year 2000 and are, except for the price for cardioversions and medication, adapted from reference 19. b These categories are aggregated to compose the cost category of professional help.

5 1546 V.E. Hagens et al. and (c) more frequent visits to the outpatient department to monitor the outcome of the cardioversion, accounted for this difference. During the first 3 months under rate control, costs for hospitalisations were slightly but significantly higher. However, after this period rhythm control generated more costs for hospitalisations. The reasons for hospital admissions in both groups consisted of a wide range of cardiovascular (e.g., evaluation of chest pain, planned diagnostic coronary angiography, admissions related to the primary endpoint) and non-cardiovascular (e.g., bone fractures, carcinoma, cataract surgery) events. Rate control remained the less expensive treatment option for persistent AF when costs on several categories were changed (Table 6). When costs for thrombosis laboratories were completely ignored, mean costs for a rate control patient were 7024 and for a rhythm control patient After the first six months of follow-up, costs of care decreased. For rhythm control, less electrical cardioversions took place (in both strategies a median of 0 electrical cardioversions). Under rate control, the number and thus the costs of cardioversions remained more or less constant during follow-up. Costs for electrical cardioversions were higher under rhythm control in every period, except during the last year of the study. At the beginning of that final year, 91 (41%) patients in the rhythm control group were still in sinus rhythm (95% of the patients used anti-arrhythmic prophylaxis). During the last year, 28 patients underwent repeated electrical cardioversion. Eventually, at the end of the study, sinus rhythm was present in 81 of the 222 subjects (37%) rhythm control patients (89% on anti-arrhythmic drugs) and in 18 of the 206 (9%) rate control patients (67% on anti-arrhythmic drugs). In all periods, costs for medication and hospital admission (except for the periods baseline until 3 months and 6 until 12 months) were significantly higher for patients under rhythm control. For both treatment strategies, costs for professional help and informal care, were high. These costs were comparable between both Table 4 Mean costs ( ) per patient between randomisation and end of study Cost category Rate control (n=206) Rhythm control (n=222) Direct medical costs Cardioversions Medication Outpatient visits Hospital admissions General practitioner visits Thrombosis laboratory Professional help Direct non-medical costs Informal care Travel costs Total Table 5 Mean costs per patient ( ) in follow-up a Baseline and 3 months 3 months and 6 months 6 months and 12 months 12 months and 24 months 24 months and end of study Rate Rhythm P b Rate Rhythm P b Rate Rhythm P b Rate Rhythm P b Rate Rhythm P b Direct M costs Cardioversions < < < < Medication < < < <0.001 Outpatient visits Hospital admissions < < < < GP visit Thrombosis lab Professional help Direct N M costs Informal care Travel costs 2 3 < Total GP, General practitioner; (N)M, (non)medical. a Patients who did not return the cost questionnaire were not analysed. Patients who did not answer the specific questions on the cost items were not analysed for that item. b Mann Whitney U test for differences between groups.

6 Rate control is more cost-effective than rhythm control 1547 groups. This reflects the relatively old study population, which is typical for patients with persistent AF. Also costs for visits to the general practitioner and thrombosis laboratory were similar in both groups. Only during the last study year patients in the rate control group generated slightly more costs, mainly due to costs for informal help ( 389). In the rhythm control group, however, costs for medication and hospital admissions were significantly higher. The direct non-medical costs were also always similar in both groups, except for travel costs but reported costs were very low. With regard to productivity losses during the first three months of follow-up, which are indirect non-medical costs, only 4 of the patients were working. Consequently, productivity losses could be ignored and, therefore, these indirect non-medical costs were not taken into account in the study. Costs of care in subgroups For daily clinical practice and for understanding the costs of care in the treatment of AF we evaluated costs in specific subgroups. In Table 7, costs in subgroups are shown, independent of the randomised strategy. Women, patients with older age, with underlying heart disease and those who reached a primary endpoint of the study generated more costs. Overall, costs were comparable between patients in sinus rhythm and AF. In patients with effective rhythm control, mean costs per patients were comparable in those who maintained sinus rhythm after 1 electrical cardioversion versus those who maintained sinus rhythm after 2 or more procedures ( 7492 versus 7671). Costs were 8573 in patients randomised to rhythm control who were in atrial fibrillation at the end of the study. Baseline characteristics of these subgroups were comparable. Discussion AFFIRM, RACE, PIAF, and STAF have shown that rate control is comparable to rhythm control concerning morbidity, mortality and quality of life in patients with persistent AF The present study showed that in patients with persistent AF, costs are lower in the rate control arm compared to the rhythm control arm. During a mean follow-up of 2.3 years, the mean costs per patient were 7386 under rate control and 8284 under rhythm control. Given the comparable morbidity and mortality, rate control is more cost effective than rhythm control. Taking into account the tendency for more endpoints under rhythm control cost-effectiveness analysis revealed that adopting a rate control strategy would save nearly per avoided primary endpoint. This is another incentive to adopt rate control therapy earlier in the course of managing persistent AF. Components of costs in the treatment of AF Varying the costs for the different cost categories also resulted in fewer costs under rate control. Also when the thrombosis laboratories were not taken into account, rate control therapy was more cost-effective. This may be important since new drugs for anti-coagulation do not necessitate coagulation monitoring. 20 The higher costs during rhythm control were predominantly caused by electrical cardioversions, more frequent visits to the outpatient department to monitor the outcome, and the use of relatively expensive antiarrhythmic drugs. This difference was primarily present during the first year after inclusion. This most likely relates to the fact that most cardioversions were performed during the first months after inclusion since recurrences of AF primarily occur early after restoration Table 6 Effect of varying 6 different cost categories on total costs ( ) and cost-effectiveness Cost category Variation Rate control Rhythm control Cost-effectiveness ratio a Cardioversions +20% % Medication +20% % Hospital admissions +20% % Thrombosis laboratory +20% % Professional help +20% % Informal care +20% % a The cost-effectiveness ratio is the difference in costs ( ) between rate and rhythm control divided by the difference in the occurrence of the primary endpoint in both groups.

7 1548 V.E. Hagens et al. Table 7 Mean costs of care ( ) in subgroups Subgroups Direct medical Direct non-medical Total Men Women Age<70 years (median) Age P 70 years (median) AF duration<445 days (median) AF duration P 445 days (median) Lone AF AF with underlying heart disease NYHA heart failure I NYHA heart failure II/III In rhythm control Sinus rhythm at study end AF at study end In rate control Sinus rhythm at study end AF at study end Endpoint reached during follow-up ,891 No endpoint reached during follow-up AF, atrial fibrillation; NYHA, New York Heart Association functional class. of sinus rhythm. 21 Thereafter, the difference in costs between rate and rhythm control declined. Previous studies investigated the cost-effectiveness during rhythm control treatment. Three studies demonstrated the supremacy of amiodarone for cost-effectiveness due to its higher efficacy for prevention of recurrences of AF In most studies, however, follow-up was rather short. Amiodarone-induced side effects are related to the cumulative dose of this drug. Therefore, outcome may be different if follow-up is extended in patients treated with amiodarone. Costs in different patient groups Subgroup analysis of the present data revealed that older patients, patients with underlying heart disease, those who reached a primary study endpoint and women caused substantially more costs. The higher costs in the first three subgroups were due to higher direct medical and direct non-medical costs and relate to a higher morbidity in these subgroups. The higher costs for women compared to men may relate to a higher incidence of the primary endpoint under rhythm control. 10 Interestingly, costs of care throughout the study were comparable for patients with sinus rhythm and AF at the end of the follow-up period. This was true for the rhythm control as well as for the rate control group. This means that costs of healthcare in these patients are largely determined by the underlying heart disease rather than the arrhythmia itself. For reduction of health care expenditures in patients with AF, it does not seem necessary to restore sinus rhythm. Limitations The present study was performed in the Netherlands and costs of health care may be different compared to other countries. However, as real costs were estimated based on true resources used, international differences regarding cost systems are irrelevant. Absolute costs may differ per country but the proportions will be comparable, and can be generalised to other countries. With a mean age of 69 years, our patients were relatively old as is usual in patients with persistent AF. 25 In the Netherlands, the age at which people reach retirement is 65 years or earlier. Therefore, costs related to time off work were not considered relevant in the present study. Furthermore, patients were asked in the questionnaire whether they had productivity losses. Only 28 patients responded to this question with 4 patients of them still working. Consequently, our cost data cannot be translated into all patients with AF. Conclusions In patients with persistent AF, costs of therapy were higher for rhythm control compared to rate control due to higher costs for electrical cardioversions, hospital admissions, and costs for medication. As the incidence of primary endpoints under rhythm control was comparable to rate control, rate control is the most costeffective treatment option for patients with persistent AF. Furthermore, the presence of sinus rhythm at study end did not lower costs of therapy.

8 Rate control is more cost-effective than rhythm control 1549 Acknowledgements The RACE study was supported by grants from the Center for Health Care Insurance (OG96-047), the Interuniversity Cardiology Institute, the Netherlands and by 3M Pharma, the Netherlands. Drs. Van Gelder and Crijns have reported receiving lecture fees from 3M Pharma. Dr. Crijns has reported serving as a consultant to Sanofi-Synthelabo and AstraZeneca. References 1. Kannel WB, Abbott RD, Savage DD et al. Epidemiologic features of chronic atrial fibrillation: the Framingham study. N Engl J Med 1982;306: Go AS, Hylek EM, Phillips KA et al. Prevalence of diagnosed atrial fibrillation in adults: national implications for rhythm management and stroke prevention: the AnTicoagulation and Risk Factors in Atrial Fibrillation (ATRIA) Study. JAMA 2001;285: Tsang TS, Petty GW, Barnes ME et al. The prevalence of atrial fibrillation in incident stroke cases and matched population controls in Rochester, Minnesota: changes over three decades. J Am Coll Cardiol 2003;42: Friberg J, Buch P, Scharling H et al. Rising rates of hospital admissions for atrial fibrillation. Epidemiology 2003;14: Stewart S, MacIntyre K, MacLeod MM et al. Trends in hospital activity, morbidity and case fatality related to atrial fibrillation in Scotland, Eur Heart J 2001;22: Stewart S, Murphy N, Walker A et al. Cost of an emerging epidemic: an economic analysis of atrial fibrillation in the UK. Heart 2004;90: Le Heuzey JY, Paziaud O, Piot O et al. Cost of care distribution in atrial fibrillation patients: the COCAF study. Am Heart J 2004;147: Van Gelder IC, Crijns HJGM, Tieleman RG et al. Chronic atrial fibrillation. Success of serial cardioversion therapy and safety of oral anticoagulation. Arch Intern Med 1996;156: Wyse DG, Waldo AL, DiMarco JP et al. A comparison of rate control and rhythm control in patients with atrial fibrillation. N Engl J Med 2002;347: Van Gelder IC, Hagens VE, Bosker HA et al. A comparison of rate control and rhythm control in patients with recurrent persistent atrial fibrillation. N Engl J Med 2002;347: Hohnloser SH, Kuck KH, Lilienthal J. Rhythm or rate control in atrial fibrillation-pharmacological Intervention in Atrial Fibrillation (PIAF): a randomised trial. Lancet 2000;356: Carlsson J, Miketic S, Windeler JJ et al. Randomized trial of ratecontrol versus rhythm-control in persistent atrial fibrillation. The Strategies of Treatment of Atrial Fibrillation (STAF) study. J Am Coll Cardiol 2003;41: Grönefeld GC, Lilienthal J, Kuck KH et al. Impact of rate versus rhythm control on quality of life in patients with persistent atrial fibrillation. Results from a prospective randomized study. Eur Heart J 2003;24: Hagens VE, Ranchor AV, Van Sonderen E et al. Effect of rate or rhythm control on quality of life in persistent atrial fibrillation. Results from the Rate Control Versus Electrical Cardioversion (RACE) study. J Am Coll Cardiol 2004;43: Gallagher MM, Camm J. Classification of atrial fibrillation. Am J Cardiol 1998;82:18N 28N. 16. Briggs AH, O Brien BJ. The death of cost-minimization analysis?. Health Econ 2001;10: Coast J, Richards SH, Peters TJ et al. Hospital at home or acute hospital care? A cost minimisation analysis. BMJ 1998;316: Oostenbrink JB, Koopmanschap MA, Rutten FFH. Guidelines for coststudies: methods and unit prices for economic evaluations in health care (in Dutch). Dutch Health Insurance Board. 19. Central medical pharmaceutical commission of the Dutch Health Insurance Board. Pharmacotherapeutic compass SPORTIF III investigators. Stroke prevention with the oral direct thrombin inhibitor ximelagatran compared with warfarin in patients with non-valvular atrial fibrillation (SPORTIF III): randomised controlled trial. Lancet 2003;362: Tieleman RG, Van Gelder IC, Crijns HJGM et al. Early recurrences of atrial fibrillation after electrical cardioversion: a result of fibrillation-induced electrical remodeling of the atria? J. Am. Coll. Cardiol. 1998;31: Catherwood E, Fitzpatrick WD, Greenberg ML et al. Cost-effectiveness of cardioversion and antiarrhythmic therapy in nonvalvular atrial fibrillation. Ann Intern Med 1999;130: Lumer GB, Roy D, Talajic M et al. Amiodarone reduces procedures and costs related to atrial fibrillation in a controlled clinical trial. Eur Heart J 2002;23: Eckman MH, Falk RH, Pauker SG. Cost-effectiveness of therapies for patients with nonvalvular atrial fibrillation. Arch Intern Med 1998;158: Levy S, Maarek M, Coumel P et al. Characterization of different subsets of atrial fibrillation in general practice in France: the ALFA study. The College of French Cardiologists. Circulation 1999;99: