For patients taking oral anticoagulants (OAC), the proportion of time spent in the therapeutic range is strongly associated with bleeding and thromboembolic risk.  Previous studies examining OAC control may not generalize because the patient population was select or international normalized ratio (INR) capture was incomplete.  The objectives of this study were to measure OAC control for an entire population of elderly people and determine patient factors associated with OAC control.

Study participants included people in Eastern Ontario without valve replacement aged 65 years or greater who were treated with warfarin between September 1, 1999 and September 1, 2000.  The study design was a retrospective cohort study using population-based administrative databases.  OAC control was measured as the proportion of days in therapeutic range (PDTR), defined as the number days with the INR between two and three divided by total number of days observation.  Linear interpolation was used to determine INR levels between measures.  Negative binomial regression was used to identify patient factors independently associated with PDTR.  Investigators also determined which factors were associated with proportion of days with a critically low (<1.5) or critically high (>/=5) INR.

7,179 people were followed for a total of 3,238 years.  15% of people were hospitalized during the study.  Overall, PDTR was 59.2% (95% CI 59.1%-59.2%).  Independent of all other significant factors, hospitalization was associated with a 15% decrease in the PDTR (rate ratio 0.85, 95% CI 0.83-0.87).  Hospitalization was also independently associated with greater proportion of time with a critically low INR (rate ratio 1.68, 95% CI 1.51-1.88) and a critically high INR (1.70, 95% CI 1.38-2.08).

Elderly people in eastern Ontario taking warfarin were therapeutic 59.2% of the time.  Independent of other patient factors, patients who are hospitalized have the greatest risk of poor anticoagulation control.  Control for anticoagulated patients who become hospitalized should be reviewed to determine if and how it could be improved.