Successful anticoagulation therapy depends on achieving therapeutic effect, measured as an increased International Normalized Ratio (INR), without incurring hemorrhagic or other complications of therapy. This balance may be difficult to achieve. Some experts have proposed that patients with mechanical heart valves aim for INR values of 2.5 to 3.5, with lower levels for other patients. Other experts recommend target INRs of 2.0 to 3.0 for all patients. Odén and Fahlé n used linked health and death records on patients attending 46 anticoagulation clinics in Sweden to estimate the INR levels associated with the lowest mortality rates.

They accessed records for more than 42,000 patients who attended anticoagulation clinics between 1990 and 1997. The mean age of the patients was 70.5 years, and 42 percent of the patients were women. The principal indications for anticoagulation were atrial fibrillation (58 percent), venous thrombosis and pulmonary embolism (25 percent), cerebrovascular conditions (22 percent), and valvular prosthesis (18 percent). Death information was available from the national registry of causes of death.

Overall mortality and death from cerebral bleeding were strongly related to increasing INR (see accompanying table). Overall, the hazard of death was lowest at an INR of 2.15. In patients with mechanical heart valves, the lowest risk of death was found to be an INR of 2.33. In other patients, the risk was lowest at an INR of 2.2. For every unit increase above an INR of 2.5, mortality increased 2.2 for all indications, 2.1 for atrial fibrillation, 2.3 for venous thrombosis and pulmonary embolism, 2.3 for stroke and transient ischemic attacks, and 2.1 for valve prosthesis. The risk of death from cerebral bleeding significantly decreased with INRs in the 1.0 to 1.5 range but significantly increased with values greater than 1.5. Deaths in patients with INR levels greater than 3.0 were more common when these levels resulted from an increase in medication rather than when this level was reached spontaneously.

The authors conclude that mortality in anticoagulated patients is lowest at INR levels of 2.2 to 2.3, and they calculate significant excess mortality at INR levels of 3.0 or more. Overall, INRs of 2.2 to 2.3 irrespective of indication appear to be associated with the lowest mortality. The authors advocate using less intensive anticoagulation therapy and maintaining patients in a narrow therapeutic window.