Guidelines on oral anticoagulation (warfarin): third edition – 2005 update

Abstract

The British Committee for Standards in Haematology (BCSH) published its third edition of Guidelines on Oral Anticoagulation in 1998 (British Committee for Standards in Haematology, 1998). Most of the recommendations made in 1998 remain unchanged and a fourth edition of the guideline is considered unnecessary at the time of writing (June 2005). However, we draw attention to those areas where new informative data have been published. As in the original guideline, the term ‘oral anticoagulant’ used in this update refers to oral vitamin K antagonists (VKA), such as warfarin. New oral non-VKA are currently being evaluated in clinical trials but are not yet licensed for use in the UK. When these drugs become available new guidance will be issued specifically for the use of those drugs. The guideline group was selected to be representative of UK-based medical experts. The drafting group met and communicated by email. MEDLINE was searched systematically for publications in English from 1998. The writing group convenor (T. Baglin) produced the draft guidelines which were subsequently revised by consensus. The guideline was reviewed by a multidisciplinary sounding board, the BCSH and the British Society for Haematology (BSH) and comments incorporated where appropriate. Criteria used to quote levels and grades of evidence are as in Appendix 3 of the Procedure for Guidelines Commissioned by the BCSH (http://www.bcshguidelines.com/process1.aspApp3). The target audience for this guideline is healthcare professionals involved in the management of patients receiving oral anticoagulant therapy. The original paragraph numbering and format used in the 1998 guideline has been retained for ease of comparison. The original Table I was modified (Table I). Treatment with oral VKA remains the treatment for choice for the majority of patients with venous thromboembolism (VTE) (van der Heijden et al, 2001). A specific exceptional patient group in which low-molecular-weight heparin therapy may be advantageous is patients with VTE complicating cancer (Meyer et al, 2002; Lee et al, 2003) (see section 3.b). Intensity of anticoagulation The Prevention of Recurrent Venous Thromboembolism (PREVENT) trial randomised patients to continue anticoagulation with a reduced target international normalised ratio (INR) of 1·5–2·0 or to take placebo, following an initial period of oral anticoagulation (Ridker et al, 2003). The median duration of treatment before randomisation was 6·5 months. Recurrent VTE occurred in the low-intensity warfarin-treated patients at a rate of 2·6/100 patient-years and in patients receiving placebo at a rate of 7·2/100 patient-years [hazard ratio (HR) 0·36, 95% confidence interval (CI) 0·19–0·67]. Bleeding rates were not significantly different but the study was not powered to test for differences in bleeding rates. The Extended Low-intensity Anticoagulation for Thrombo-Embolism (ELATE) investigators randomised patients who had been treated with a target INR of 2·5 for at least 3 months to continue anticoagulation with a reduced target INR of 1·5–1·9 or continue treatment with a target of 2·5 (Kearon et al, 2003). Recurrent VTE occurred in the low-intensity warfarin-treated patients at a rate of 1·9/100 patient-years and in patients receiving conventional treatment with a target INR of 2·5 at a rate of 0·7/100 patient-years (HR 2·8, 95% CI 1·1–7·0). Bleeding rates were not significantly different but the bleeding rate in patients on conventional intensity (0·9/100 patient-years) was lower than the anticipated rate that was used to power the study (3/100 patient-years). When comparing thrombosis rates in the trials it is evident that there is a dose–response effect. Combining PREVENT and ELATE the cumulative thrombosis rate at 4 years was 2·5% with a target of 2·5, 7·5% with a target INR range of 1·5–1·9/2·0 and 20% without warfarin (Ridker, 2004). Therefore a target INR of 2·5 is the most effective. As there was no difference in bleeding between targets of 2·5 and 1·75 (target range 1·5–1·9/2·0) a target INR of 2·5 should remain the default target for patients requiring long-term oral anticoagulation for prevention of VTE. It has been suggested that the findings do give some support for a lower intensity of therapy in patients at very high risk of bleeding (Ridker, 2004). It is possible that in high bleeding-risk patients, the bleeding rates would be different for a target of 2·5 and 1·75. Patients can be stratified for bleeding risk, for example prospectively with the Outpatient Bleeding Risk Index (Landefield & Beyth, 1993) or following a period of observation on treatment. On an exceptional basis, a target INR of 1·75 might be adopted for those patients at risk of recurrent VTE but who have a very high risk of bleeding that might ordinarily result in treatment being stopped (grade C, level IV). Such patients might be those who have already suffered a major bleed and in whom the risks for bleeding have not altered or those patients with repeated episodes of overanticoagulation, for example an INR greater than 8 requiring reversal with vitamin K or coagulation factor replacement on more than one occasion. A target INR of 2·5 is recommended for long-term oral anticoagulant (VKA) therapy for secondary prevention of VTE (grade A, level 1b). A meta-analysis of long versus short duration anticoagulant therapy after a first episode of VTE indicated a relative risk of recurrence of 0·6 (95% CI 0·45 to 0·79) for a duration treatment of 3 months or greater compared to 6 weeks or less (Pinede et al, 2000). Two recently published trials present data that may influence decisions regarding duration of oral anticoagulation after an episode of VTE. The Short term Oral anticoagulation for a First Acute Secondary Thrombosis (SOFAST) investigators assessed duration of anticoagulation following a first episode of deep vein thrombosis (DVT) or pulmonary embolus (PE) provoked by a transient risk factor in a double-blind study in which patients who had completed 1 month of anticoagulant therapy were randomly assigned to continue warfarin or placebo for an additional 2 months (Kearon et al, 2004). The aim was to determine if the duration of treatment could be reduced without an increase in the rate of recurrent VTE during 11 months of follow-up. Recurrence rates were not significantly different, 6·0% with placebo and 3·7% with warfarin (P = 0·5). There were no major bleeds in either group. A meta-analysis of four other studies combined with the SOFAST study was also performed; in combination, the results of the five studies suggested that shortening the duration of anticoagulant therapy to 4 or 6 weeks more than doubled the frequency of recurrent VTE within a year of diagnosis (odds ratio 2·9, 95% CI 1·2–6·9). The authors’ conclusion was that duration of anticoagulant therapy for VTE provoked by a transient risk factor should not be reduced from 3 months to 1 month. The Durée Optimale du Traitement AntiVitamines K (DOTAVK) study was an open-label, randomised trial comparing 3 with 6 months anticoagulation after proximal deep vein thrombosis (DVT) and comparing 6 weeks with 3 months after calf vein thrombosis (Pinede et al, 2001). Recurrence rates and bleeding rates were not different in patients receiving short-duration as compared to long-duration therapy. Equivalence was observed in both intention-to-treat and per-protocol analyses. Recurrence rates were lower in patients with temporary, as compared with permanent, risk factors but duration of therapy did not influence recurrence rates in either group. The authors, who also performed the meta-analysis quoted above (Pinede et al, 2000), concluded that 3 months of treatment may be sufficient for patients with temporary risk factors and a low risk of recurrence. Anticoagulation for 1 month is inadequate treatment after an episode of VTE (grade A, level 1b). At least 6 weeks anticoagulation is recommended after calf vein thrombosis (grade A, level 1b) and at least 3 months after proximal DVT or PE (grade A, level 1b). For patients with temporary risk factors and a low risk of recurrence 3 months of treatment may be sufficient. For patients with idiopathic VTE or permanent risk factors at least 6 months anticoagulation is recommended. Several studies have now been reported in which patients have been randomised to different durations of anticoagulation. In theses studies recurrence rates whilst on anticoagulant treatment with a target INR of 2·5 have been similar in patients with and without heritable thrombophilic defects (level II). Similarly, recurrence rates in studies investigating the influence of thrombophilia testing on recurrence of VTE have demonstrated low rates of recurrence whilst on treatment with a target INR of 2·5 in patients with and without heritable thrombophilia (level II). The duration of anticoagulation was not influenced by the presence of laboratory evidence of thrombophilia in most cases (British Committee for Standards in Haematology, 2001). Extended duration of anticoagulation for patients considered to be at excessive risk of recurrent VTE will generally be based on clinical criteria, such as recurrent idiopathic events and family history. Such decisions will generally be the same for patients with and without identifiable heritable defects. Two randomised trials have compared a target INR of 2·5 (range 2·0–3·0) to a target greater than 3·0 [range 3·1–4·0 (Crowther et al, 2003) or 3·0–4·5 (Finazzi et al, 2005)]. Odds ratios for recurrent thrombosis in the high intensity compared to low-intensity groups were 3·1 (95% CI 0·6–15·0) (Crowther et al, 2003) and 1·97 (95% CI 0·49–7·89) (Finazzi et al, 2005). Major bleeding rates were not different. Based on these studies both groups of authors concluded that a target INR of 2·5 was sufficient for treatment of patients with thrombosis in association with antiphospholipid syndrome. Both studies included patients with either venous or arterial thrombosis but the majority had venous thrombosis. Therefore a target of 2·5 is recommended for patients with VTE. There are insufficient data to make an evidence-based recommendation for patients with antiphospholipid syndrome and arterial thrombosis but a higher target of 3·5 is often used and is reasonable as long as the bleeding risk at the higher intensity of anticoagulation is taken into consideration. A target of 3·5 is also recommended for patients who suffer recurrence of VTE whilst on warfarin with an INR between 2·0 and 3·0. A target INR of 2·5 is recommended for patients with DVT or PE associated with antiphospholipid syndrome (grade A, level Ib). The management of iliofemoral venous thrombosis in injecting drug users is problematic because of poor venous access, non-compliance with prescribed treatment, ongoing intravenous drug use and co-existent sepsis. It is unlikely that a randomised trial of treatment with warfarin versus low molecular weight heparin (LMWH) alone would be practical in this patient group. Because of the great variation in response to warfarin between and within patients it is not appropriate to use oral VKA when monitoring of the patient is not possible or probable for whatever reason. There are retrospective data suggesting that treatment with LMWH alone results in a satisfactory clinical outcome (Mackenzie et al, 2000). Treatment with LMWH is an alternative to oral anticoagulation in patients with VTE secondary to intravenous drug use (grade C, level IV). British Committee for Standards in Haematology (1998) recommended a target INR of 2·5 for patients awaiting cardioversion (grade C). This recommendation was based on extrapolation of the efficacy of a target INR of 2·5 in patients with non-rheumatic atrial fibrillation. A recent study confirmed the need for anticoagulation in patients awaiting cardioversion (Gallagher et al, 2002). No thromboembolic events occurred in 779 cardioversion attempts when the INR was 2·5 or greater before cardioversion, nine events occurred in 756 cardioversion attempts when the INR was less than 2·5 (seven patients) or not measured (two patients). The authors concluded that the INR should be 2·5 or greater at the time of cardioversion. A commentary accompanying the Gallagher publication recommended that this study alone was insufficient evidence to change practice (Olshansky, 2002). In this commentary, Olshanky presented data from his own unit, showing a very low risk of events (0/532) when the INR was greater than 2·0 for three consecutive weeks prior to cardioversion. Increasing the target INR to 3·0 might reduce the stroke rate slightly and it is unlikely that this benefit would be offset by an increased incidence of bleeding over a short period of time. In the UK it is common practice to measure the INR on the day of the cardioversion and postpone the procedure if it is less than 2·0. This results in cancellation of as many as 25% of procedures. The main benefit of adopting a higher target INR, for example 3·0, would be the avoidance of a cancellation due to an INR <2·0 on the day of the procedure. Already, some centres adopt a target INR of 3·0 for the 4 weeks prior to the procedure and a target INR of 2·5 afterwards. We consider such an approach relatively safe and useful (grade C recommendation) although we can make no absolute recommendation that it should be adopted. A target INR of 2·5 is recommended for 3 weeks before and 4 weeks after cardioversion (grade B, level III). To minimise cardioversion cancellations due to low INRs on the day of the procedure a higher target INR, e.g. 3·0, can be used prior to the procedure. The frequency of thromboembolism is lower with modern valves than first generation valves but oral anticoagulation with VKA is still required. As the intensity of anticoagulation is a major determinant of bleeding risk the Seventh American College of Chest Physicians (ACCP) Conference on Antithrombotic and Thrombolytic Therapy refined the recommended target INRs for patients with modern valves et al, 2004). It is that studies are still to determine risk factors patients with and of the level of anticoagulation and the level of anticoagulation at which there is now sufficient level evidence from which to make recommendations for target INRs when on and are (Table II). There is on or valves and a target INR of 3·5 is still recommended. It is that bleeding or with may a in target INR to 3·0, or in some of these such valves are now being and this is sufficient is not available a target INR of 3·0 is recommended for valves in the and 3·5 in the For patients in whom and are specific target INRs are recommended (Table II). a target INR of 3·0 is recommended for valves in the and 3·5 in the Patients requiring vein who had risk factors for were randomised to or warfarin to a target INR of 2·5 in a study et al, 1998). patients heparin before arterial and patients randomised to warfarin an of heparin for to the time ratio at the bleeding was significantly higher in the anticoagulation group was higher in the anticoagulation group and the cumulative rates were higher with a significantly higher cumulative rate = The results of the study may have been due in to the of heparin in the first drugs remain first for secondary long-term anticoagulation is to patients at high risk of vein a target INR of 2·5 is recommended (grade B, level A recent meta-analysis et al, indicated that warfarin the rate of or stroke more than alone although the risk of bleeding is The studies included in the meta-analysis were before was there was no It is that patients at high risk and low bleeding risk will benefit from combined therapy compared to alone although the for patients, and in the influence of on patient management decisions regarding oral anticoagulation remain to be oral anticoagulation with a vitamin K is in to a target INR of 2·5 is recommended (grade A, level I). oral anticoagulant therapy is prescribed a target INR of 2·5 is recommended (grade A, level I). A retrospective of patients with for a median of 6 years a cumulative thrombosis rate of et al, 2003). The risk of thrombosis in patients with was compared with in those with had been to patients with and a greater than There were no events in patients with who warfarin to a target INR of 2·5 the cumulative thrombosis rate in patients with not receiving warfarin was The incidence of major was Based on the available data it was not possible for the authors to often should be or to to in anticoagulation with a target INR of 2·5 is recommended for patients with and a greater than (grade B, level III). Anticoagulation can also be considered for patients with and less than on additional risk factors for thrombosis and bleeding (grade C, level IV). Two randomised studies have demonstrated of LMWH over oral anticoagulant therapy with warfarin in patients with The combined outcome measure of major bleeding or recurrent VTE was more in patients treated with warfarin (target INR compared to in the study reported by et relative risk (95% CI = No bleeding was observed in patients receiving whilst patients in the warfarin group of bleeding (P = one of these patients was The Thrombosis compared 6 months of to after 1 to oral anticoagulation with a vitamin K warfarin (target INR The outcome measure was recurrent VTE during the study Secondary were clinical bleeding and The ratio for recurrent VTE in patients receiving was (95% CI = et al, 2003). Major bleeding was not significantly different (P = In a patients treated with with without at the time of diagnosis of VTE had et al, 2005). is generally to LMWH for treatment of VTE in patients with cancer (grade A, level Ib). was assessed in requiring anticoagulation for atrial et al, 2004). Patients were on 3 of warfarin for 1 and by INR day of patients had an INR greater than 2 and had a by day and by day The day 8 INR was of 11 patients had an INR greater than 4 and no patient suffered a or bleeding in the first month. alternative in which patients 2 warfarin for 2 weeks was to the from the INR et al, 1998). In the one patient had an INR greater than 3·0 in the first 2 weeks and five had INRs greater than 4 at some time after than of INRs were between 2·0 and 3·0 after A has been evaluated in with atrial fibrillation. The of the INR on and 8 & 1998). to treated with the conventional recommended in the guideline (British Committee for Standards in Haematology, this in a lower INR during the first of therapy = and INRs greater than to anticoagulation was similar with the a more of for versus = A warfarin was compared with the in in an randomised study et al, 2000). In both patients were on day 1 and were by INR to day For INRs the of warfarin was 2·5 lower on day 2 and lower on day 3 with the trial The time to anticoagulation was with the trial but patients had INRs in the first 8 This difference was for patients more than The to the within 1 was for both A of other have also et al, et al, et al, et al, 2003). et and et suggested that a a INR as as a with less excessive anticoagulation. et that if was for the first 2 a more response was into range et suggested a lower should be used in those above years of For who do not anticoagulation a is safe and anticoagulation in the majority of patients within weeks (grade B, level This to and bleeding associated with For patients requiring of oral anticoagulation that with or a by may be to that with repeated in patient e.g. the years of those with or and those at high risk of bleeding (grade B, level of a after oral anticoagulant therapy has in as to treatment should be stopped or a in some patients following of oral anticoagulant of the of et al, & In many patients this is the result of a that may have to the anticoagulant treatment. A of versus on laboratory of coagulation has been This was for and factor C but not 1 after randomisation when one group of patients had stopped and one group had taken of et al, However, the was 1 after treatment of of there was no difference between the groups 1 after the treatment to clinical risk, retrospective studies have results & studies have not indicated a & (level and thromboembolic rates are low when anticoagulation is in patients with valves & (level III). this evidence there is no need for of anticoagulant therapy. Oral anticoagulant therapy can be when the duration of therapy is completed (grade B, level A of the and efficacy of management for patients on oral requiring did not randomised trials & 2003). an of thromboembolic events were in patients 95% CI 95% CI to management thromboembolic rates were for of oral for without of for with of heparin and for with of Major bleeding was of oral anticoagulation for the following and and and or with or without data in to valves and not the for anticoagulation in of the management the of this 2003). A for the intensity of anticoagulation on the day of INR has been reported et al, 2003). When oral anticoagulation is stopped should be to low heparin recommendations remain there is a very high risk of thromboembolism anticoagulation should be in for Anticoagulation not need to be stopped for for patients in INR of anticoagulation with vitamin K is more with intravenous than oral et al, 2001). In the original guideline an of of vitamin K or was recommended for patients with major in to factor replacement therapy with either a factor or We now consider in patients with major reversal with intravenous vitamin K is A of either or is recommended. and reversal of is more with a factor than with et al, et al, 2001). vitamin K should be if reversal is to be et al, 2002). In a study the of of was less when the INR was less than 2·0 at the time of the or was reduced to less than 2·0 within of with a factor was associated with a lower incidence of as compared with reversal with et al, 2003). of anticoagulation in patients with major bleeding of a factor in to when this is available (grade B, level and of intravenous than oral vitamin K (grade B, level of patients are testing for monitoring long-term oral anticoagulant therapy. patients are also In an of a the majority of patients who were a et al, 2004). patients have from without with or by clinical anticoagulant treatment. patients should be to and with clinical and after for It is that will in patients to before to New guidelines on and are in For either or Patients should with or without within a The same of management as in should be Patients should be assessed for patients considered to management should and with or without as appropriate. and should be reviewed and at for both and clinical should of of INRs in range and the incidence of overanticoagulation, bleeding and Standards for as in Table of the British Committee for Standards in Haematology (1998) guideline with the that it is recommended that be altered to a of of The range is taken as INR of the The for the change is that the time in range is now considered to be more than the of INRs within target et al, 1993) and clinical has that this target is The need for of outcome data is of of patients from to management of anticoagulation during with drug of duration of therapy and remain areas of practice where of often to be a anticoagulant therapy to be as a risk with of patients being to both and of treatment. were included in a recent of of as that the of additional The is currently to practice recommendations for to be issued to the in of the authors have a of the and in these guidelines is to be and at the time of to the authors and the British Society for Haematology the for the of these