Skip navigation
Link to Back issues listing | Back Issue Listing with content Index | Subject Index

Optimum INR level?


Large, comprehensive linkage studies can provide useful information about various aspects of healthcare. Many come from Sweden, and the latest [1] provides much information about the INR levels and mortality.


The computerised records of 42,000 patients attending 46 anticoagulation clinics in the years 1990-1997 formed the basis of the study. Anticoagulation was being provided mainly for atrial fibrillation (58%), venous thrombosis and pulmonary embolism (25%), stroke and transient ischaemic attacks (22%), and valve prostheses (18%), or more than one of these indications.

Patient deaths were identified from a registry of causes of death, that included 99% of all deaths in Sweden. Contributing to this analysis were 3,553 deaths.

There were 1.25 million INR measurements. Each patient was followed from the time of an INR measurement to either the next INR, death, or for seven weeks. After this data were censored until a new INR measurement was performed.


The average age of patients at the start of anticoagulation treatment was 71 years. The outcomes were mortality per 1,000 patient years, and death from bleeding per 1,000 patient years. Ascertainment for the latter as a cause of death was not complete because the fact of bleeding could not be confirmed in all patients.

All cause mortality was strongly related to the INR value. The lowest death rates were with INR values of 2.0 to 2.9 (Figure 1). Mortality was higher at INR values below 2.0. At INR values above 5.0, mortality was about 800 per 1,000 patient years.

Figure 1: All cause mortality versus INR after 1.25 million INR measurements in 42,000 patients

Death caused by bleeding was also strongly related to INR. At INR values above 3.0 the rate increased rapidly (Figure 2).

Figure 2: Death from bleeding versus INR after 1.25 million INR measurements in 42,000 patients


The authors concluded that mortality was lowest at INR values between 2.0 and 2.3. Mortality was not just from cerebral haemorrhage or bleeding, and other factors contributed more to all cause mortality. The window for optimal anticoagulation therapy was very narrow.


  1. A Odén, M Fahlén. Oral anticoagulation and risk of death: a medical record linkage study. BMJ 2002 325: 1073-1075.
previous or next story in this issue