Unfortunately, few data are available to help clinicians identify the appropriate time interval between the discontinuation of warfarin therapy and a surgical procedure. Most published recommendations are not referenced and appear to be based primarily on personal experience [5-7]. The primary objective of our study was to measure the decrease in the international normalized ratio (INR) over time after discontinuation of warfarin therapy.

Methods

Top Methods Results Discussion Author & Article Info References

Patients

Our study was approved by the committee for research involving humans at the University of California, Davis, School of Medicine. The risk for thrombosis during several days without anticoagulant therapy was judged to be acceptably low for patients with 1) deep venous thrombosis after 12 weeks of treatment; 2) atrial fibrillation with no valvular heart disease, stroke, or systemic embolization; and 3) congestive cardiomyopathy without heart failure, previous stroke, or systemic embolization. Patients with a history of widely fluctuating warfarin requirements and those with a complex dosing regimen that precluded simplification were excluded.

Protocol

Patients received a fixed daily dose of warfarin at 5:00 p.m. For example, a patient who had been receiving 5 mg/d 3 times weekly and 7 mg/d 4 times weekly at 8:00 am (43 mg/wk) received 6 mg/d each evening (42 mg/wk). After at least 1 week of this regimen, each patient was seen midday on a Monday (day 1), and his or her INR was measured. Patients were then instructed to discontinue warfarin therapy and to return 48 hours (day 3), 96 hours (day 5), and 168 hours (day 8) later for repeated INR measurements.

Five patients whose INR had a wide range in the rate of decrease were restudied at least 1 month after restarting daily warfarin therapy. These patients had INR measurements at 8:00 a.m. and 4:00 p.m. for 5 consecutive days after discontinuation of warfarin therapy.

Measurements

Laboratory INR measurements were done on an automated photo-optical coagulometer (MLA-700, Medical Laboratory Automation, Inc., Mount Vernon, New York). The thromboplastin used for all but the first 8 patients studied was a recombinant human tissue factor (Innovin, Baxter Diagnostics, Deerfield, Illinois) with an international sensitivity index of 0.92 (lot TFS 12). Thromboplastin C (Baxter Diagnostics), which has an international sensitivity index of 2.98 (lot 627), was used for the first 8 patients.

Calculations and Statistics

The half-life of the INR was calculated after transforming INR values to the natural logarithm of (INR –1) and measuring the slope of the linear regression versus time in hours after discontinuation of warfarin therapy. The quantity (INR –1) was used because the minimum value for INR is 1.0. Because the INR did not decrease until after the second measurement on day 1, the first INR measurement on day 1 was excluded. The theoretical time at which the INR began to decrease from the steady-state level was calculated by using data from day 2 to day 5 and by extrapolating the linear regression equation back to the steady-state level equal to the INR in the afternoon of day 1. Means are expressed ±SD.

Relations between various clinical variables and the percentage decrease in the INR between day 1 and day 3 were evaluated using stepwise multiple linear regression [8]. The percentage decrease in the INR between day 1 and day 3 was the dependent variable; variables analyzed were age, weight, height, sex, dose of warfarin, initial INR, and time between the day 1 INR and the day 3 INR. Regression analysis was done using a software program (Solo, BMDP statistical software, Los Angeles, California) on a personal computer.

Results

Top Methods Results Discussion Author & Article Info References

Thirty-two of 44 potential candidates agreed to participate in the study; 22 completed the protocol. Twelve were men and 10 were women; their average age was 55 years (range, 25 to 81 years); their mean weight was 85.5 kg; and their mean height was 172 cm. Nine had atrial fibrillation, 10 had venous thromboembolism, and 3 had cardiomyopathy; their mean warfarin dose was 5.7 mg/d (range, 1.5 to 11.0 mg/d). Their mean initial INR was 2.6; all INRs were between 2.0 and 3.0 except for one that was 1.95 and one that was 3.8. Figure 1 shows INR values as a function of time after the last dose of warfarin. On day 3, a mean of 66.5 ±2 hours after the last dose of warfarin, 10 of 22 patients (55%) had an INR greater than 1.5, whereas only 2 of 22 (9%) had an INR less than 1.2. On day 5, an average of 115 ±7 hours after discontinuation of warfarin therapy, all but 5 patients had an INR less than 1.2, and on day 8, with the exception of two INRs that were 1.2 and 1.23, all INRs were less than 1.2. Using linear regression analysis, we found that the only covariate showing a significant (P = 0.005) independent association with the percentage decrease in the INR between day 1 and day 3 was age (regression coefficient = –6.8%±2%/2 days per decade of age; R² = 0.34).

View larger version (9K): [in this window] [in a new window]   Figure 1. Decrease in the international normalized ratio (INR) over time after discontinuation of warfarin therapy. The measured INR is depicted as a function of time after the last dose of warfarin.

Discussion

Top Methods Results Discussion Author & Article Info References

No previous study has analyzed the decrease in the INR or prothrombin time over time after discontinuation of long-term warfarin therapy in either normal persons or patients at risk for thromboembolism in whom warfarin therapy is discontinued before noncardiac surgery [9-12]. Review articles that discuss the perioperative management of patients receiving warfarin therapy make recommendations based only on clinical experience [5-7].

Our findings indicate that the INR decreases exponentially beginning 24 to 36 hours after the last dose of warfarin. In the 5 patients studied in detail, the half-life of the INR varied considerably, ranging from approximately 0.5 to 1.2 days (12 to 29 hours). Interpatient variation in both the time until exponential decrease begins and the actual rate of decrease are likely contributors to the wide range of INR values (1.1 to 2.16) that we noted in our total cohort approximately 65 hours after the last dose of warfarin.

By analyzing the percentage change in the INR from the steady-state (day 1) value to the day 3 value, which was taken 65 hours after discontinuation of warfarin therapy, we found that age was the only clinical variable significantly associated with percentage decrease. For each 10-year increase in age, the fall in the INR decreased by approximately 7%. This is not surprising; it is known that older persons eliminate warfarin more slowly than do younger persons [13].

Our study did not address the following question: What preoperative INR value is associated with a low incidence of major bleeding complications? Tinker and Tarhan [11] report that almost all major hemorrhages during or after major noncardiac surgery occurred in patients with excessively high prothrombin times, which, assuming that a typical thromboplastin was used, corresponds to an INR greater than 1.46. Assuming that an INR of less than 1.5 is associated with an acceptable risk for perioperative bleeding, our data suggest that almost all patients with an initial INR between 2.0 and 3.0 will have an INR less than 1.5 115 hours (about 4.8 days) after discontinuation of warfarin therapy. Delaying surgery for this length of time is particularly prudent in older patients who eliminate warfarin more slowly and in patients at high risk for a bleeding complication. In patients who must continue to receive anticoagulation therapy except during surgery, administration of a small quantity of vitamin K followed by intravenous heparin may be necessary [5, 10]. Finally, unless the risk for bleeding is negligible, the INR should always be measured before proceeding with surgery.