[indices] Adverse events in patients with warfarin

["Leesette Turner" <hamis@iafrica.com>]

Re: Meta-analysis: About half the serious adverse events on warfarin due to poor control

Cross posted as fair use from National Electronic Library of Medicines.

       
      Source CMAJ 2007; 176: 1589-94 
      Resource Links Abstract, full text available free  
      Abstract Around half of the adverse events in patients taking warfarin- whether bleeds or thromboemboli - are related to poor anticoagulant control, according to a meta-analysis. The Canadian authors of the study note that anticoagulant control in patients taking oral anticoagulants is often poor, and that better control has been shown to improve both bleeding and thromboembolic event rates. Reliable values for the proportions of these events that are related to poor control will allow better judgement of the effect of measures to improve it: this study aimed to provide robust estimates using published data. 

      The authors carried out a fairly comprehensive literature search for studies that reported rates of major haemorrhages or thrombotic events in patients taking oral anticoagulants, and where the INR at the time of the event was reported, and the majority of patients in the study had INRs within generally accepted ranges. From each study, they extracted the total numbers of bleeds and thromboemboli that could be related to particular INR ranges. Events were then grouped into three INR ranges - above, below, and within therapeutic range; for each study, they then calculated the proportions of major bleeds that occurred when INR was above the therapeutic range, and of thromboemboli that occurred when below it. Standard techniques were then used to pool the values into overall proportions. 

      The literature search produced 362 potentially relevant articles, of which 45 were eligible for analysis (n=71,065). The most frequent reason for exclusion was that a study did not assign events to an INR range (n=288). Of the eligible studies, 23 reported both bleeds and thromboemboli, 14 bleeds only, and 8 thromboemboli only. About two-thirds were carried out in community practice, and about two-thirds were of moderate or good quality. Nearly a third used approximate INRs based on values taken 2 to 30 days before the event. 

      Overall, 44% (95% CI 39% to 49%) of haemorrhages occurred while the patient's INR was above the therapeutic range, and 48% (95% CI 41% to 55%) of thromboemboli occurred at INRs below therapeutic range. For those studies that reported time spent proportions of time spent within or outside therapeutic range, the time spent below therapeutic range was significantly associated with the proportion of thromboemboli occurring below range; such a relationship was not found for bleeding events occurring above range, however. 

      The authors conclude that in patients taking oral anticoagulants, about half the serious haemorrhages and half the thromboembolic events will be associated with INR below or above the therapeutic range respectively. They also note the converse; that half the events occurred within range. Although their study has limitations, it suggests that improved control of oral anticoagulation therapy could stop about half the serious haemorrhages and thromboemboli that occur during use. They suggest that further research using patient-level data is needed to clarify the issue further. 


[Note from moderator: Can anyone share any personal experiences in this regard or similar stories about poor monitoring of other drugs? Carolyn]