Quality of oral anticoagulant control and treatment in Sweden

OBJECTIVES: To define laboratory-dependent and clinical factors that negatively influence the precision and safety of oral anticoagulation and to determine whether this creates differences in clinical outcome between the participating hospitals. DESIGN: Laboratory. The clinical chemistry laboratories of participating hospitals performed prothrombin time tests on blinded, standardized plasma samples on six occasions. Clinical. Patients with a first or second episode of venous thromboembolism were randomized to different durations of oral anticoagulation; the target was an international normalized ratio (INR) of 2.0-2.85. SETTING: Multicentre study at the departments of medicine of 16 Swedish hospitals. SUBJECTS: In total, 1124 patients with venous thromboembolism were followed for 600 patient-years of oral anticoagulation. Main exclusion criteria were previously known malignancy or venous ulcer, known congenital deficiency of an inhibitor of coagulation and unwillingness to participate. MAIN OUTCOME MEASURES: Laboratory. Interlaboratory variation was measured with coefficient of variation. Clinical. End-points were recurrent venous thromboembolism and haemorrhages requiring hospitalization or treatment with blood products or vitamin K. RESULTS: Laboratory. The interlaboratory variation in prothrombin time analyses was 11.3% at a mean INR of 3.8. No difference was detected between laboratories using the two prevalent thromboplastin reagents in Sweden or between those using the Behnk Coagulator and ACL instruments. INR results. Seventy-five per cent of the INR values were > or = 2.0, and 58% were within the target range. The time spent within the target range was between 57 and 74% at the worst and best hospital, respectively. Referral of patients to satellite clinics and fear of treating patients living in distant villages too intensively were factors that decreased the number of effectively anticoagulated patients. The percentage of patients effectively anticoagulated was lower amongst those < 50 than those > or = 50 years of age and also lower during the 1st year than during the 2nd and 3rd. Clinical events. These were eight objectively verified events of recurrent venous thromboembolism [1.3 in 100 patient-years or 0.7%; 95% confidence limits (CL): 0.2 1.2]. Seventeen haemorrhagic events occurred, corresponding to 2.8 per 100 patient-years or 1.5% (CL: 0.8, 2.2); two fatal haemorrhages corresponding to 0.3 per 100 patient-years or 0.2% (CL: 0.0, 0.4). The difference in the incidence of these complications between hospitals with > 60 and < or = 60% of patients effectively anticoagulated did not reach statistical significance. Patients with haemorrhagic complications were not older than the rest. CONCLUSIONS: The performance of the laboratories was acceptable. Clinically important differences between the hospitals were not observed. The incidences of thromboembolic and haemorrhagic complications were low, even in comparison with other randomized trials concerning venous thromboembolism. However, it might be possible to reduce the risk of haemorrhage further with increased centralization and improved education of patients as well as medical staff, and perhaps to reduce the risk of recurrent venous thromboembolism by aiming at a more intensive range of anticoagulation during the 1st month.