Risk of haemorrhage associated with long term anticoagulant therapy.

This literature review was conducted to determine: (a) the rate of bleeding (major, minor and fatal) during long term oral anticoagulant therapy (greater than 4 weeks) in various disorders (ischaemic cerebrovascular disease, prosthetic cardiac valves, chronic atrial fibrillation, ischaemic heart disease and venous thrombosis); and (b) the clinical and laboratory risk factors which predispose such patients to bleeding. Using strictly defined methodological criteria, 167 studies were evaluated and classified into 1 of 5 categories based on the strength of the study design, with level I (randomised trials) representing studies which provided the most reliable information and level V (cases series) the least reliable. The risk of bleeding was substantial, and was most marked in patients with ischaemic cerebrovascular disease (29%), ischaemic heart disease (19%) and venous thromboembolism (23%). Major bleeding in venous thrombosis and cerebrovascular disease was frequently associated with an underlying risk factor. In venous thromboembolism these coexisting conditions (cancer, recent surgery and paraplegia) were also predisposing factors for thrombosis. In cerebrovascular disease major bleeding was almost always intracerebral, possibly because of associated hypertension or the cerebrovascular disease per se. We were unable to determine whether bleeding events were concentrated soon after commencing anticoagulant therapy. Haemorrhagic episodes frequently occurred when the prothrombin time (or thrombotest) was within the targeted therapeutic range, but the relationship between bleeding and the level of anticoagulant therapy was properly evaluated in only 1 study (in venous thrombosis) which demonstrated that the risk of bleeding was reduced by using a less intense anticoagulant regimen. In conclusion, the risk of bleeding during oral anticoagulant therapy is substantial. Our analysis was limited by the lack of concise reporting of clinical and laboratory information and we would suggest that future clinical studies report these in greater detail.