Gains and losses of warfarin therapy as performed in an anticoagulation clinic

Transcription

1 Journal of Internal Medicine 2006; 259: doi: /j x Gains and losses of warfarin therapy as performed in an anticoagulation clinic A. M. NJAASTAD 1,U.ABILDGAARD 1 &J.F.LASSEN 2 From the 1 Thrombosis Clinic, Department of Medicine, Aker University Hospital, Oslo, Norway; and 2 Department of Cardiology B, Skejby Hospital, Århus University Hospital, Århus, Denmark Abstract. Njaastad AM, Abildgaard U, Lassen JF (Aker University Hospital, Oslo, Norway; and Skejby Hospital, Århus University Hospital, Århus, Denmark). Gains and losses of warfarin therapy as performed in an anticoagulation clinic. J Intern Med 2006; 259: Background. Knowledge of the net benefit of warfarin therapy in routine care is needed to define realistic management recommendations, but lack of randomized controls precludes conventional risk benefit analysis. Objective. Assess risk and benefit of routine warfarin therapy in an anticoagulation clinic. Design. Retrospective observational analysis. Patients. A total of 1435 outpatients on warfarin for a total of 1613 patient years, treated to prevent the target events recurrent venous thromboembolism (VTE) or myocardial infarction (MI), and stroke in patients with atrial fibrillation (AF) or mechanical heart valves. Measurements. Major bleeding and thromboembolic (TE) events and all deaths. Calculations. Expected annual target event rates without warfarin were from published data. Differences between combined major events observed with warfarin, and expected without warfarin were calculated. Results. In the total material, annual rates were 3.0% major TE events, 1.1% major bleeding events, 0.12% fatal bleeding, and a benefit/risk ratio of 3.8. The net gain, expressed in reduced combined bleeding and target TE annual event rate, was 9.9% in secondary prophylaxis in AF, 4.4% in VTE patients, 2.7% in post-mi patients, 2.4% in primary prophylaxis in AF and 0.6 in patients with mechanical heart valves. The apparent benefit/risk ratio was 3.9 in VTE patients, 5.8 in AF patients and 1.1 in patients with mechanical heart valves. Conclusion. Net effects of prolonged warfarin therapy in patients with VTE and AF performed in an anticoagulation clinic have an acceptable risk/ benefit ratio, comparable with what has been obtained in elective clinical trials. Keywords: atrial fibrillation, myocardial infarction, risk benefit analysis, stroke, venous thromboembolism, warfarin. Introduction Prospective randomized trials have demonstrated the efficacy of warfarin therapy in preventing recurrence of venous thromboembolism (VTE) [1] and myocardial infarction (MI) [2], as well as reducing the risk of stroke in patients with atrial fibrillation (AF) [3] or prosthetic mechanical heart valves [4]. The increased risk of bleeding is the main problem of antithrombotic therapy. Rates of major bleeding in patients treated with warfarin have ranged 1 17 per 100 patient years in randomized trials [5]. The benefit in reducing thromboembolic (TE) events, however, was several-fold larger. Based on the results warfarin therapy is widely used. National sales statistics suggest that about 1% of the population in the Nordic countries currently is on warfarin therapy. In the routine situation, rates of major bleeding may range 2 6 per 100 patient years, and fatal bleeding per 100 patient years [6 8]. The introduction of anticoagulation clinics, the expression of anticoagulant effect in International 296 Ó 2006 Blackwell Publishing Ltd

2 NET GAIN OF WARFARIN THERAPY 297 Normalized Ratio (INR) values with better defined therapeutic ranges, and the use of computer software to assist dosing, have contributed to lower incidences of bleeding [5]. Rates of major bleeding events [9 11], and fatal bleeding events per 100 patient years [10, 11] have been reported from anticoagulation clinics. Increased bleeding risk is clearly related to excessive anticoagulation [5]. The risk is higher in advanced age [11 13], is increased by patient comorbidity [5, 13, 14] and may be higher during the initial 3 months [5, 6]. It is possible that with the recent tendency to a lower risk of bleeding, the antithrombotic effect of warfarin therapy may outweigh the hazard even in elderly patients with more comorbidity than seen in elective studies, but this is not well documented. There is thus a need for assessing the net effects of warfarin therapy in routine treatment. The present report concerns major bleeding and TE events in consecutive patients treated with warfarin in the routine of an anticoagulation clinic. We wished to assess the net clinical benefit in these patients. Lacking a randomized control group made a strict risk benefit analysis impossible. Instead, the rates of major adverse events observed were compared with the rates that might be expected without warfarin therapy, as assessed from data on control patients in randomized studies, preferably when reported in contemporary meta-analyses. Care was taken to interpret these data with caution so as not to exaggerate the assessed gains of therapy. In patients with VTE, the risk of recurrence after cessation of anticoagulant therapy depends strongly both on the length of warfarin therapy period and on patient characteristics [1, 15 23]. Estimates of the risk of recurrent VTE after warfarin therapy for 6 months were used for comparison. In patients with AF, the incidence of a primary ischaemic stroke in controls is fairly uniform [3, 24] and similar to the estimate in a population study [25]. Stroke rates are much higher in AF patients who have sustained an ischaemic stroke [26]. In patients with MI, the risk of recurrent infarction depends on patient characteristics. Our comparison was based on the rate observed in controls observed in the WARIS study [27]. In patients with prosthetic mechanical heart valves, the risk of stroke depends greatly on type and position of the mechanical valve. Our comparison with results in historical controls was based on a comprehensive meta-analysis [28] but should still be regarded as tentative. We interpreted the extracted literature as to provide reliable and consistent control data which should permit a semi-quantitative assessment of the therapeutic gain in patients with VTE and with AF. In patients with MI and particularly in patients with mechanical heart valves, control data were less consistent which called for an even more cautious assessment. Methods Monitoring routines The patients were recruited at discharge from the hospital. The majority of patients had their first visit at the anticoagulation clinic 1 3 weeks after warfarin therapy had been initiated. A few patients, mostly with AF who showed a stable response to warfarin were transferred to primary care. At each visit the patient saw a physician who adjusted the warfarin dose and decided the time interval until next visit. Warfarin was used by about 99% of the patients, but was changed to phenindione in those with allergic side-effects. INR was determined with a capillary method using the Thrombotest reagent, ISI around 1.02 from Nycomed, Oslo, Norway. All INR values, dosages, comorbid conditions, possible events and drug interactions were recorded in the patients AC charts. INR targets The target range was INR in patients with VTE and AF, and in post-mi patients and patients with mechanical heart valves. Diagnostic procedures Deep venous thrombosis (DVT) was verified by ultrasonography or venography. Pulmonary embolism was confirmed by CT angiography of the lungs or a ventilation-perfusion scan. Ischaemic stroke was confirmed by CT in patients with residual neurological deficits beyond 24 h. Transient ischaemic attack (TIA) was defined as neurological deficits of presumably vascular origin lasting <24 h and without recent changes in cerebral CT. TIAs were not classified as major events. Acute MI was confirmed by elevation of myocardial enzymes and typical changes in ECG.

3 298 A.M.NJAASTADet al. Definitions of events Bleeding was defined as major if it was associated with at least one of the following: death; intracranial, retroperitoneal, intraocular or intra-articular bleeding; a decrease in haemoglobin level of at least 20 g L )1 ; need for transfusion of two or more units of blood; or need for surgical or medical interventions [11]. TE events were defined as major if they were fatal, resulted in acute MI, stroke, or necessitated an emergency surgical procedure. Recurrent venous TE episodes were also classified as major events. An evaluation of a causal role of warfarin therapy was based on all clinical data available, and classified as one of the four alternatives: definite, probable, possible and unlikely [29]. In the fatal cases who had sustained major bleeding, the causal role of warfarin therapy on the fatal outcome was also evaluated by Steinar Madsen, The Norwegian Medicines Agency. Time in therapeutic range was calculated according to Rosendaal et al. [30]. Data sources The INR values and intervals between consultations were traced from the AC charts. All data from these charts were transferred to a database. The diagnosis registry of Aker University Hospital covering all admissions for these patients during , i.e. including admissions occurring up to 1 month after the study period, was checked for all patients. For all admissions in which the main or subsidiary diagnoses might represent or allude to bleeding or thrombotic episodes, the complete hospital records were analysed. Based on these records, we thus retrospectively registered clinical, laboratory and autopsy data regarding all bleeding and TE events which had led to or occurred during hospitalization. Events causing hospitalizations outside Aker University Hospital had been noted in the AC chart and hospital reports were obtained. Deaths were traced from the hospital records and the National Death Registry. Estimates of the risk of TE events in patients not treated with warfarin In patients with VTE, the risk of recurrence depends on the length of warfarin therapy as well as patient characteristics. A short treatment time may result in 20 30% recurrences [20], and is only sufficient if a transient risk factor is present [1, 16, 17]. In other patient categories, an elevated risk persists longer and treatment for 6 months has been considered optimal for most patients [16]. In the present patient material, many patients were treated beyond 6 months, and we sought for data on the risk of recurrence in this period. In two studies, in which patients with high risk thrombophilia and cancer were excluded, a continuing annual risk of recurrence was 4 5% [18 20]. In patients with idiopathic VTE, cancer, or high risk thrombophilia, the risk of recurrence is considerably higher [21 23], and suggested to be above 10% per year [16]. More than half of our patients had conditions associated with an elevated risk of recurrence. We based our assessment on an expected rate of recurrent VTE of 7%, but included a calculation assuming a risk of 4%. In patients with AF included in primary prevention trials and not receiving antithrombotic therapy, the annual risk of ischaemic stroke averaged 4.6% [24]. A stroke rate of 12.3% was reported in AF patients who have sustained a previous ischaemic stroke [26]. In patients with MI, there are few data on the frequency of recurrence in the more recent reports. In the WARIS study, in which patients were monitored for an average of 37 months, the annual frequency of recurrent infarction was 6.6% in the controls [27], and this value was used. In patient with mechanical heart valves, the reported frequency of stroke varies considerably, but is very high in some series [31]. A comprehensive metaanalysis indicated annual rates of ischaemic stroke of 4% in the absence of antithrombotic therapy [28]. The benefit/risk ratio was the number of prevented major TE events divided by the number of observed major bleeding events. The number of prevented major TE events was obtained by subtracting the number of observed TE events on warfarin, from the number expected without warfarin. Statistics Differences between groups in categorized variables were analysed with chi-square test using the SPSS v program (SPSS Inc., Chicago, IL, USA). Analyses were two-sided, and differences were considered statistically significant at the level <0.05.

4 NET GAIN OF WARFARIN THERAPY 299 Results Demography of the population The population consists of all outpatients treated with vitamin K antagonist (VKA) at the Anticoagulation Clinic, Aker University Hospital, Oslo, during 5 years and 2 months (from 21 September 1991 to 1 December 1996), with the indications VTE, AF, MI and mechanical heart valve prostheses. The analysis was limited to these four indications, which comprised 1435 patients (Table 1). The mean age was 67.4 years. The median age was 71 years, range years, highest in AF (75 years), lowest in MI patients (68 years). About 56.4% of the patients were males. An additional 226 patients (294 patient years) were also treated for the indications like cerebrovascular disease, peripheral arterial disease and congestive heart failure, but these were not included in the present study as we were unable to find consistent data in recent literature that allowed for a calculation of a possible benefit of treatment with warfarin. Medication with aspirin was not allowed. The exception was one patient with mechanical aortic valve prosthesis who was on additional aspirin medication because she had previously suffered from probable embolic episodes despite warfarin with INR in therapeutic range. Examination of charts and records for all patients with major bleeding failed to disclose the use of aspirin in any of them. Patients with VTE (11%) had a verified and symptomatic pulmonary embolism. DVT was proximal in 75%. Malignant disease was present in 11% of the patients, and their VTE was treated until recurrence, or cessation of warfarin because of general deterioration or death. High-risk thrombophilia (deficiency of either antithrombin, protein C or protein S) was diagnosed in 18% and antiphospholipid antibody syndrome in 4%, and warfarin was administered on an infinite basis. About 22% had a recurrent VTE and were treated for 12 months. The remaining patients were treated for 6 months, except patients with distal DVT and transient risk factor who were treated for 3 months. Of the total treatment years, malignancy and high-risk thrombophilia accounted for about 50%, and spontaneous VTE for about 10%. Of patients treated more than 6 months, about 45% had cancer or thrombophilia. About 78% of the patients with AF were on primary and 22% on secondary prophylaxis. Patients who had sustained MI were referred to AVK therapy if an elevated risk of thrombosis was suspected because of cardiac aneurysm, large anterior wall infarction, a history of thrombosis or suspected refractarity to aspirin. Amongst 151 patients with mechanical heart valve prosthesis, 106 were with aortic and 45 with mitral mechanical valve. Anticoagulation Time in defined therapeutic ranges (TTR) was highest for patients with mechanical heart valves (92.1%), followed by VTE (74.3%), IHD (73.5%), AF (71.5%). The mean interval between visits was 21 days. The mean treatment time was 1.15 years, longest for patients with mechanical heart valves (1.95 years). The longer treatment time, and the wider INR range may explain the higher TTR in the mechanical heart valve group. Fatalities There were 35 deaths in the population during the observation period, 28 from causes unrelated to TE or bleeding. Five were caused by TE and two by Table 1 Number of patients, patient years, major TE and bleeding events, according to indication for warfarin Indication No. of patients Patient years Bleeding events TE events Bleeding event rates TE event rates VTE (1) 15 (1) 1.5 (0.2) 3.1 (0.2) AF (1) 10 (3) 0.8 (0.2) 2.1 (0.6) MI (0) 16 (0) 0.0 (0.0) 4.4 (0.0) MHV (0) 7 (1) 2.4 (0.0) 2.4 (0.3) Total (2) 48 (5) 1.12 (0.12) 2.98 (0.31) TE events include all types of major TE events irrespective of indication. The numbers of fatal events, and fatal event rates, are listed in parentheses. VTE, venous thromboembolism; AF, atrial fibrillation; MI, myocardial infarction; MHV, mechanical heart valves; TE, thromboembolic.

5 300 A.M.NJAASTADet al. major bleeding. The annual mortality in the patient population was 0.022, which is not significantly lower than which is the calculated rate in a sex-matched (57% males) Norwegian population aged years in 1996 (National Bureau of Statistics). Major TE events Forty-eight major TE events occurred in 43 patients, 28 males and 15 females. Their mean age was 68.8 years, median age 73 years, similar to those in the total population. There was no significant difference in the frequency of major TE events in patients above and below 75 years, nor in the frequency of TE events in males and females (chisquare analysis). Thirty-five TE target events occurred in patients on warfarin (Table 5). Of these, MI was fatal in one and stroke fatal in three patients. In addition, there were 13 TE events of a different type than the target indication (Table 2). Of these, one patient who sustained a femoral artery occlusion died. Major bleeding Eighteen major bleeding events, two of them fatal, occurred in 15 patients (11 were males and four females; Table 3). Their mean age was 66.0 years, median age 70 years, similar to those in the total population. There was no significant difference in the frequency of major bleeding in patients above or below 75 years, nor in the frequency of bleeding events in males and females (chi-square analysis). International Normalized Ratio on admission was above therapeutic range in six of 18 events (33%). Table 2 Rates of the major TE event types in relation to indication for warfarin Indication MI Stroke VTE Art TE VTE AF 2.1 Primary prophylaxis 1.7 Secondary prophylaxis 3.1 MI Mechanical heart valves Total Rates are per 100 patient years. VTE, venous thromboembolism; AF, atrial fibrillation; MI, myocardial infarction; TE, thromboembolic; Art TE, acute TE occlusion of the femoral artery. Table 3 Major bleeding events VTE AF One patient with pneumonia and peritonitis died. As necropsy revealed peritoneal bleeding, warfarin therapy was interpreted as possibly contributing to death. In two patients who died 2 and 3 weeks after cessation of warfarin with carcinomatosis and with endocarditis and renal failure respectively, bleeding was interpreted as not contributing to the fatal outcome. Patients with mechanical prosthetic heart valves had the highest incidence of major bleeding (Table 2). INR was above therapeutic range in four of the seven events. Two patients had cancer which caused local bleeding. How many target TE end-points were possibly prevented? Warfarin apparently prevented 27 recurrences of VTE (Table 4). Using an alternative expected rate of 4%, as relevant in VTE patients without elevated risk, the calculated minimum gain is reduced to 15 recurrences. Six of seven recurrent VTE events occurred during the initial 6 months of therapy. Accordingly, calculations restricted to the events observed beyond treatment for 6 months would have suggested an additional increase in benefit of six events. For the total material, the assessed gain was a prevention of 68 major TE events (Table 5). Estimation of the net clinical effect Mechanical heart valves MI Total Bleeding events Patients Malignancies Fatal bleeding 1 a 1 b INR >ther. range Number of events and patients affected, listed according to indication for warfarin. INR, International Normalized Ratio; VTE, venous thromboembolism; AF, atrial fibrillation; MI, myocardial infarction. a Sepsis, haemorrhagic peritonitis, INR 2.8. b Intracranial bleeding, INR 4.8. The observed 18 major bleeding events reduces the assumed benefit to a net benefit of 50 major events prevented during the 1613 treatment years (Table 5). Benefit/risk are 3.8 for the total patient material, 3.9 in patients with VTE, 5.8 in patients with AF and 1.1 in patients with mechanical

6 NET GAIN OF WARFARIN THERAPY 301 Table 4 Assessment of the number of TE events prevented Target TE events Indication for warfarin Patient years treated Observed on warfarin Expected without warfarin Prevented TE events VTE, recurrences MI, recurrences AF, primary stroke prophylaxis AF, secondary stroke prophylaxis Mechanical heart valves, stroke prophylaxis Total TE events that were the target of therapy in each group are shown. The numbers of expected events were calculated from the number of patients years treated, and published annual rates in controls without antithrombotic therapy, as explained in Methods. The expected annual rates used were: 7.0% recurrent VTE (cf. Methods), 4.6% for a first stroke in AF [24], 12.3% for a recurrent stroke in AF [26], 6.6% recurrent MI [27], 4.0% stroke in mechanical heart valves [28]. VTE, venous thromboembolism; AF, atrial fibrillation; MI, myocardial infarction; TE, thromboembolic. Table 5 Assessed net clinical gain of warfarin versus no antithrombotic therapy, expressed in rates per 100 patient years Indication Observed on warfarin Expected without warfarin TE Bleeding TE Bleeding Net clinical gain VTE, recurrence MI, recurrence AF, primary prophylaxis AF, secondary prophylaxis Mechanical heart valves Total The net gain is the difference between the sum of expected rates of target TE and major bleeding events observed on VKA, and the sum of these event rates expected without therapy. Expected rates of major bleeding in patients without antithrombotic therapy are from the publications used for assessment of expected major TE events [19, 26, 27]. We observed no major bleeding in our MI patients on warfarin. Major bleeding rate in post-mi patients without warfarin was reported as 0.2 [27], but in our calculation we used the value 0.0, as it is not logical that warfarin reduces the rate of bleeding. Rates of major bleeding in control patients with mechanical heart valves are not available [28] and the mean of the rates for the other indications was used. VTE, venous thromboembolism; AF, atrial fibrillation; MI, myocardial infarction; TE, thromboembolic. prosthetic heart valves. In the MI patients no major bleeding was observed, and a benefit/risk ratio could not be calculated. The net gain was highest in secondary stroke prophylaxis in AF patients, followed by VTE patients and primary stroke prophylaxis in AF patients (Table 5). A high rate of major bleeding reduced the net effect in patients with mechanical heart valves (Table 5). Discussion Warfarin treatment caused 18 major bleeding events in the present patient material. Our assessments suggest that the treatment may have prevented about 68 major TE events, or 3.8 times as many major TE events as the major bleeding events. The use of rates reported in controls from published trials implies that the numbers of the prevented TE events could only be crudely assessed. The apparent gains were largest in patients with AF and VTE. Our results further support the use of prolonged warfarin therapy in patients with high risk of recurrent VTE. These conclusions are supported by direct as well as indirect lines of evidence. The direct evidence is, first, that the observed rate of major bleeding on warfarin therapy, 1.1% per year, is similar or lower to rates observed in most trials, and lower than the % per year reported in cohort studies [33]. Secondly, the

7 302 A.M.NJAASTADet al. observed rates of major TE events on warfarin (Table 5) are not higher than reported in the majority of clinical trials [1, 3, 17, 21, 24, 27]. Thirdly, time in target INR ranges (TTR) was obtained in as much as 72 92%. In a recent review, 28 of 35 studies had reported TTR values below 70% [32]. The indirect evidence is the comparison with the major event rates that should be expected in these patients if they had not received warfarin therapy. The expected rates were mainly based on data from published results in control groups performed in the same period as the present study. We sought to interpret these data carefully in order to avoid unrealistic assessments. In calculating gains, we gave the same importance to major bleeding as to major TE events. Although this may be relevant regarding VTE patients, an ischaemic stroke or a recurrent MI may be more harmful to the patient s quality of life than will most major bleeding episodes. The risk of recurrent VTE varies considerably between patient categories. Our calculation was based on an expected rate of 7% annual recurrence rate after treatment for 6 months. The high proportion of high-risk patients in our material could mean that we underestimated the gain. The same is true if we had only based the calculations on the single recurrent TE event observed beyond 6 months of treatment. We chose to assess the probable gain soberly to counteract the potential errors of comparing with published rates. However, our data may support that the gain in prolonging treatment of TE beyond 6 months may be restricted to patients who have an increased risk of recurrence, in agreement with current recommendations [1, 16, 17]. In a prospective study on a large group of patients with AF, Go et al. [34] showed that primary warfarin prophylaxis reduced the annual stroke rate by 53%. It was concluded that warfarin prophylaxis may be as effective and safe in clinical practice as in elective trials. Our result supports this conclusion. The observed annual stroke rate in our patients on primary warfarin prophylaxis was 63% lower than the expected rate without warfarin (1.7% vs. 4.6%, Table 5). The patient group with prosthetic heart valves contained fewer patients than the other groups. Bleeding in cancer tumours, and in patients with excessive anticoagulation resulted in relative high rate of major bleeding. Pengo et al. [35] reported bleeding from a gastric carcinoma in one of 15 patienets with major bleeding. We are not aware of other reports of cancer-induced bleeding in this patient category, nor indications that patients with malignancy have been excluded from reported materials. We compared the results observed on warfarin therapy with those who might be expected with aspirin in the prevention of stroke and recurrent MI. Apparently, the advantage of using warfarin was moderate, not only in patients with mechanical heart valves, but also in primary prophylaxis in AF, and in post-mi patients (data not shown). Any clear increase in major bleeding rates in these patient categories would endanger the net benefit of warfarin therapy. Our findings and calculations thus emphasize that a low rate of bleeding is essential to obtain a clear benefit of warfarin treatment. This is particularly relevant for post-mi patients without conditions that increase the risk of thrombosis, and which may have an acceptable effect of antiplatelet therapy [36]. Limitations of the study The retrospective analysis design involves a risk for missing end-points. We cannot exclude that patients may have sustained major nonfatal events when staying outside our area and that these events were not reported in our charts and case records. The use of published events rates in nontreated patients implies that our assessments should be regarded as semi-quantitative. The small number of patients in the subgroups makes it difficult to know for sure if one group benefits more from warfarin therapy than others. We believe that our results in the mechanical heart valve group underestimated the general benefit of warfarin therapy, which has been documented in other reports [28, 31]. Conclusion Net effect of prolonged warfarin therapy in patients with VTE and in patients with AF performed in an anticoagulation clinic have an acceptable benefit/ risk ratio, comparable with what has been obtained in clinical trials. The low rates of major adverse events, and a high proportion of INR values in desired ranges (TTR values), in the present study,

8 NET GAIN OF WARFARIN THERAPY 303 confirm earlier reports from anticoagulation clinics. The high TTR values may reflect relatively shorter intervals between patients visits. More importantly, the advantage of anticoagulation clinics probably reflects greater experience and dedication to this specialized service. Conflict of interest statement No conflict of interest was declared. Acknowledgements Jakob Hjort and Finn Strekerud developed the database. Jakob Hjort performed the calculations. Steinar Madsen evaluated the possible role of warfarin therapy in fatal cases who had suffered from major bleeding. Nycomed, Norway, gave financial support to the transference of AC chart data to the database. Anne Mette Njaastad received financial support from The Quality Assurance Fund of The Norwegian Medical Association. References 1 Büller HR, Agnelli G, Hull RD, Hyers TH, Prins MH, Raskob GE. Antithrombotic therapy for venous thromboembolic disease: the seventh ACCP conference on antithrombotic and thrombolytic therapy. Chest 2004; 126: 338S 409S. 2 Hackam DG, Anand SS, Yusuf S. Oral anticoagulant therapy in patients with coronary artery disease. Semin Vasc Med 2003; 3: Singer DE, Albers GW, Dalen JE, Go AS, Halpern JL, Manning WJ. Antithrombotic therapy in atrial fibrillation: the seventh ACCP conference on antithrombotic and thrombolytic therapy. Chest 2004; 126: 429S 56S. 4 Cannegieter SC, Rosendaal FR, Wintzen AR et al. Optimal oral anticoagulant therapy in patients with mechanical heart valves. N Engl Med J 1995; 333: Levine MN, Raskob G, Beyth RJ, Kearon C, Shulman S. Hemorrhagic complications of anticoagulant treatment. the seventh ACCP conference on antithrombotic and thrombolytic therapy. Chest 2004; 126: 311S 37S. 6 Landefeld CS, Beyth RJ. Anticoagulant-related bleeding: clinical epidemiology, prediction, and prevention. Am J Med 1993; 95: Hurlen M, Erikssen J, Smith P, Arnesen H, Rollag H. Comparison of bleeding complications of warfarin and warfarin plus acetylsalicylic acid: a study in 3166 outpatients. J Intern Med 1994; 236: Steffensen FH, Kristensen K, Ejlersen E, Dahlerup JF, Sor HT. Major haemorrhagic complications during oral anticoagulant therapy in a Danish population-based cohort. J Intern Med 1997; 242: Chiquette E, Amato MG, Bussey HI. Comparison of an anticoagulation clinic with usual medical care: anticoagulation control, patient outcomes, health care costs. Arch Intern Med 1998; 168: Poli D, Antonucci E, Lombardi A et al. Low rate of bleeding and thrombotic complications of oral anticoagulant therapy independent of age in the real-practice of an anticoagulation policlinic. Blood Coagul Fibrinolysis 2003; 14: Palareti G, Leali N, Coccheri S et al. Bleeding complications of oral anticoagulant treatment: an inception-cohort, prospective collaborative study (ISCOAT). Lancet 1996; 348: Fang MC, Chang Y, Hylek EM et al. Advanced age, anticoagulaqtion intensity, and risk for intracranial hemorrhage among patients taking warfarin for atrial fibrillation. Ann Int Med 2004; 141: Launbjerg J, Egeblad H, Heaf J, Nielsen NH, Fugleholm AM, Ladefoged K. Bleeding complications to oral anticoagulant therapy: multivariate analysis of 1010 treatment years in 551 outpatients. J Intern Med 1991; 229: Landefeld CS, McGuire E, III, Rosenblatt MW. A bleeding risk index for estimating the probability of major bleeding in hospitalized patients starting anticoagulant therapy. Am J Med 1990; 89: Pinede L, Duhaut P, Cucherat M, Ninet J, Pasquier J, Boissel JP. Comparison of long versus short duration of anticoagulant therapy after a first episode of venous thromboembolism: a meta-analysis of randomised, controlled trials. J Intern Med 2000; 247: Bates SM, Ginsberg JS. Treatment of deep-vein thrombosis. N Engl J Med 2004; 351: Prins MH, Hutten BA, Koopman MMW, Büller HR. Long-term treatment of venous thromboembolic disease. Thromb Haemost 1999; 82: Schulman S, Rhedin AS, Lindmarker P et al. A comparison of six weeks with six months of oral anticoagulation therapy after a first episode of venous thromboembolism. N Engl J Med 1995; 332: Schulman S, Granquist S, Holmström M et al. The duration of oral anticoagulation therapy after a second episode of venous throemboembolism. N Engl J Med 1997; 336: Schulman S. Duration of anticoagulants in acute or recurrent venous thromboembolism. Curr Opin Pulm Med 2000; 6: Kearon C, Gent M, Hirsh J et al. A comparison of three months of anticoagulation with extended anticoagulation for a first episode of idiopathic venous thromboembolism. N Engl J Med 1999; 340: Prandoni P, Lensin AW, Cogo A et al. The long-term clinical course of acute deep venous thrombosis. Ann Intern Med 1996; 125: Prandoni P, lensing AWA, Piccoli A et al. Recurrent venous thromboembolism and bleeding complications during anticoagulant treatment in patients with cancer and venous thrombosis. Blood 2002; 100: Hart RG, Benavente O, McBride R et al. Antithrombotic therapy to prevent stroke in patients with atrial fibrillation: a meta-analysis. Ann Intern Med 1999; 131: Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation: a major contributor to stroke in the elderly: The Framingham Heart Study. Arch Intern Med 1987; 147: EAFT (European Atrial Fibrillation Trial) study group. Secondary prevention in non-rheumatic atrial fibrillation after transient ischemic attack or minor stroke. Lancet 1993; 342:

9 304 A.M.NJAASTADet al. 27 Smith P, Arnesen H, Holme I. The effect of warfarin on mortality and reinfarction after myocardial infarction. N Engl J Med 1990; 323: Cannegieter SC, Rosendaal FR, Briët E. Thromboembolic and bleeding complications in patients with mechanical heart valve prostheses. Circulation 1994; 89: Hallas J, Harvald B, Gram LF et al. Drug related hospital admissions: the role of definitions and intensity of data collection, and the possibility of prevention. J Int Med 1990; 228: Rosendaal FR, Cannegieter SC, van den Meer FJ, Briet E. A method to determine the optimal intensity of oral anticoagulant therapy. Thromb Haemost 1993; 69: Stein PD, Alpert JS, Copeland J, Dalen JE, Goldman S, Turpie AGG. Antithrombotic therapy in patients with mechanical and biological prosthetic heart valves. Chest 1992; 102 (Suppl.): 445S 54S. 32 Ansell J, Hirsh J, Poller L, Bussey H, Jacobson A, Hylek E. The pharmacology and management of the vitamin K antagonists. The seventh ACCP conference on antithrombotic and thrombolytic therapy. Chest 2004; 126: 204S 33S. 33 Schulman S. Care of patients receiving long-term anticoagulant therapy. N Engl J Med 2003; 349: Go AS, Hylek EM, Phillips KA et al. Effectiveness and safety of warfarin to prevent thromboembolism among patients with atrial fibrillation: how well do results of randomised trials apply to clinical practice? The ATRIA study. JAMA 2003; 290: Pengo V, Barbero F, Banzato A et al. A comparison of moderate with moderate-high intensity if oral anticoagulant treatment in patients with mechanical heart valve prosthesis. Thromb Haemost 1997; 77: Anand SS, Yusuf S. Oral anticoagulant therapy in patients with coronary heart disease: a meta-analysis. JAMA 1999; 282: Correspondence: Ulrich Abildgaard, Department of Medicine, Aker University Hospital, N-0514 Oslo, Norway. (fax: ; ulrich.abildgaard@ioks.uio.no).