The cohort consisted of all patients receiving continuous oral anticoagulant therapy with either warfarin or coumarin derivatives on October 15, 2002, registered at the Outpatient Anticoagulant Clinic at the University Hospital of Umeå. It is the only hospital in the reception area. A small number of patients treated by general practitioners (n = 57) were not included in the study. In most cases monitoring by general practitioners is due to long distances in our catchment area. Due to the same reason, it may be estimated that the number of patients treated by other hospitals is negligible. During the time of the study, only one patient living in our area used a self-monitoring device, and he was supervised by the Outpatient Anticoagulant Clinic, and thus is included in our study.
All patients in the register with a treatment time defined as "indefinite" were considered to be on continuous oral anticoagulant therapy. The patient records from relevant clinics were examined. Personal data, start date of anticoagulant treatment, and all thromboembolic events in the patient's history that came to our knowledge were recorded. If previously unknown events were found in personal communication with the patient, these events were also recorded. For all patients without mechanical heart valve prosthesis, the reason for the ongoing anticoagulant treatment was recorded.
Mechanical valve prosthesis, dilated cardiomyopathia, or venous thrombosis, in combination with one of the following coagulopathies, were defined as absolute indications for continuous warfarin treatment: antithrombin deficiency, protein C deficiency, protein S deficiency, homozygote for F V Leiden (FV 1691 AA), homozygote for the prothrombin mutation (FII 20210 AA), heterozygote for both APC-resistance (FV 1691 GA) and prothrombin mutation (FII 20210 GA). If no problems with the treatment were registered by the nurses at the Outpatient Anticoagulant Clinic, these patients were not subjected to any further investigation in this study.
Patients with atrial fibrillation were divided into two groups with low or high risk for cardioembolic events. A patient with atrial fibrillation at low risk was defined one who had none of the following: age more than 60 years, previous cerebrovascular event, dilated cardiomyopathia, mitral stenosis, marked mitral insufficiency, or marked enlargement of the left atrium. All other patients with atrial fibrillation were defined as high risk. High risk was considered a legitimate reason for continuous anticoagulant treatment as long as the risk for bleeding complications was considered low. Low risk patients were invited for a medical examination, and discontinuation of anticoagulant treatment was considered, with a possible switch to other medications.
Patients with recurrent venous thromboses (three or more) without known risk factors were offered a medical examination and the presence of coagulopathy was examined. Anticoagulant treatment of these patients was continued if there were no contraindications.
All other legitimate indications were considered as relative indications, and they were weighed against the risk of bleeding complications. Risk factors and possible contraindications were: documented serious bleeding complications, as well as factors increasing risk of bleeding such as problems with compliance, fluctuating INR, balance problem with documented slip or fall accidents, dementia, liver failure and high age (>85 years of age).
Patients with unclear indication or with known risk factors were invited for further investigation. The indication for treatment was re-evaluated after laboratory tests and medical examination was performed. If no valid indication was found, or if the risk was found to be too high, anticoagulant treatment was discontinued or replaced with other appropriate medication whenever possible.
INR was determined with STA® – SPA 50 kit (Diagnostica Stago, Asnieres-sur-Seine, France) on a Sysmex® CA-6000 automatic coagulation instrument (Sysmex Corporation, Kobe, Japan). In mid-November INR values for the subjects were recorded. The first INR value dated on or after October 15, 2002 was recorded. In absence of such a value, the first INR value before this date was used. All values, except one, were determined within two months before this date.
Demographic data was collected from the SCB, Statistics Sweden website, http://www.ssd.scb.se/databaser/makro/start.asp, accessed January 6, 2005.
Information about warfarin use in Sweden was retrieved from the database PharmX supplied by Läkemedelsinformation AB.
Data was collected using a data sheet produced with SPSS Data Entry Builder 3.0, and SPSS™ version 12.0 (SPSS inc, Chicago, IL, USA) was used for statistical analysis. For comparison between groups, one-way ANOVA with Bonferoni post-hoc testing was used. A p-value <0.05 was considered statistically significant. All statistical tests and corresponding p-values were two-sided.
Ethical permission was granted by the local ethical committee of Umeå University.