Variation in Warfarin Use at Hospital Discharge After Isolated Bioprosthetic Mitral Valve Replacement: An Analysis of the Society of Thoracic Surgeons Adult Cardiac Surgery Database.

BACKGROUND Anticoagulation with warfarin following bioprosthetic mitral valve replacement (BMVR) is recommended by multiple practice guidelines. We assessed practice variability and patient characteristics associated with warfarin prescription following BMVR. METHODS We analyzed 7,637 patients in the Society of Thoracic Surgeons Database (January 1, 2008 to June 30, 2011) who were discharged following isolated primary nonemergent BMVR. Patients requiring preoperative warfarin, those with preoperative atrial fibrillation, or those with a contraindication to warfarin were excluded. The association between patient, hospital, and surgeon characteristics and warfarin prescription were evaluated. RESULTS Fifty-eight percent of this cohort (median age, 66 years; female sex, 58.7%) was prescribed warfarin. Patients receiving warfarin were older (67 vs 65 years; P < .0001), were less likely to have had preoperative stroke (9.3% vs 12.1%; P < .001), CHF (51.4% vs 54.1%; P < .02), or dialysis (4.9% vs 9.0%; P < 0.001), and had a longer postoperative length of stay (8.0 vs 7.0 days; P < 0.01). Warfarin was prescribed less often for patients with postoperative GI events (44.4% vs 55.6%; P < .001) but more often for patients with postoperative myocardial infarction (75.8% vs 24.2%; P < .001) or new atrial fibrillation (68% vs 32%; P < .001) and those requiring blood transfusions intraoperatively (55.7% vs 44.3%; P < .001) or postoperatively (57% vs 43%; P < .03). Similar rates of warfarin prescription were observed in patients requiring reoperation for bleeding (54.9% vs 45.1%; P = .20) and those with postoperative stroke (53.6 % vs 46.4 %; P = .30). After adjusting for patient characteristics, significant surgeon and hospital variation in warfarin prescription at hospitals was observed. CONCLUSIONS Although patient characteristics and postoperative events may be associated with the prescription of warfarin following BMVR, substantial surgeon and hospital variability remains. This variability largely ignores the established practice guidelines and warrants further study to define the optimal anticoagulation strategy in patients undergoing BMVR.