Twenty years' single-center experience with mechanical heart valves: a critical review of anticoagulation policy.

BACKGROUND AND AIM OF THE STUDY: Since January 1990, a variety of mechanical valves (St. Jude Medical, CarboMedics, ATS Medical) have been implanted routinely at the authors' institution. The study aim was to analyze, retrospectively, the 20-year clinical results of those mechanical valves, and to challenge the anticoagulation policy employed over the years.

METHODS: Between January 1990 and December 2008, a total of 2,108 mechanical valves was inserted into 1,887 consecutive patients (1,346 aortic, 725 mitral, 27 tricuspid, 10 pulmonary). The mean age of the patients was 63 +/- 13.2 years, and the majority (61%) were males. Preoperatively, 71% the patients were in NYHA class > or = III (average 3.01). The most frequent comorbidities included: atrial fibrillation (n = 594), coronary disease (n = 567) and diabetes (n = 398). The follow up (99% complete) totaled 13,721 patient-years (pt-yr), and ranged from 12 to 241 months (average 84 months).

RESULTS: In-hospital mortality was 5.2% (n = 98, 14 valve-related). Of the 629 late deaths, the majority were cardiac (n = 276). Survival (Kaplan-Meier estimation) was significantly better for aortic valve patients compared to mitral or multiple valve replacement (Mantel-Cox, p < 0.0001). The overall linearized incidences (as %/pt-yr) were: valve thrombosis 0.31, thromboembolism 1.08, and bleeding 0.91. However, as repeated events occurred in several patients, the hazard function was not constant. Multivariate analysis (Cox regression model) showed age > 70 years (p < 0.0001), NYHA class > or = III (p < 0.0001), non-sinus rhythm (p = 0.001), concomitant coronary artery bypass grafting (p = 0.008) and higher International Normalized Ratio (INR) values (p = 0.013) as significant risk factors for death, with a trend for redo operations (p = 0.052). Multivariate analysis found variable INR, non-sinus rhythm and NYHA class > II as significant risk factors for thromboembolism, while long-acting coumadin and NYHA class > II were significant risk factors for bleeding.

CONCLUSION: This 20-year experience demonstrated excellent clinical outcomes for patients with mechanical prostheses, with no valve structural failure and an acceptable incidence of adverse events. INR values between 2-2.5 for aortic valve patients, and 3-3.5 for mitral valve patients, yielded the fewest major adverse events.