Medline ® Abstract for Reference 71

BACKGROUND: Whether clinicians should decrease the warfarin dose in response to a mild, asymptomatic elevation in the international normalized ratio (INR) is unknown.

OBJECTIVES: The study objectives were as follows: (1) to evaluate the safety of an anticoagulation service (ACS) policy advocating that the warfarin dose not be changed for isolated, asymptomatic INRs of<or = 3.4; (2) to compare the dosing strategies of an ACS and primary care providers (PCPs); and (3) to quantify the relationship between reduction of the warfarin dose and the subsequent fall in the INR.

DESIGN AND SETTING: Randomized controlled study of health maintenance organization outpatients who were receiving warfarin.

PATIENTS: We identified 231 patients with a target INR of 2.5 and an isolated, asymptomatic INR between 3.2 and 3.4. Our ACS monitored 103 of the patients; PCPs monitored the remaining 128 patients.

MEASUREMENTS: From all 231 patients, we obtained INRs and warfarin dosing history. From the 103 ACS enrollees, we also recorded adverse events.

RESULTS: One ACS patient had epistaxis in the 30 days after the elevated INR. Twenty-three percent of ACS enrollees and 47% of PCP patients reduced their warfarin dose (p<0.001). The median follow-up INRs were similar in both cohorts: 2.7 in the ACS enrollees and 2.6 in the PCP patients. However, in a subgroup analysis of 190 patients who presented with an INR of 3.2 or 3.3, ACS enrollees were more likely to have a follow-up INR in the range of 2 to 3 (p = 0.03). The median follow-up INR was 2.7 in 148 patients who maintained their warfarin dose, 2.5 in 77 patients who decreased their dose by 1 to 20%, and 1.7 in 6 patients who decreased their dose by 21 to 43% (p<0.001).

CONCLUSIONS: These findings support maintaining the same warfarin dose in asymptomatic patients with an INR of<or = 3.3, and reducing the dose for patients who have a greater INR or an increased risk of hemorrhage. Warfarin dose reductions>20% should be avoided for mildly elevated INRs.