Thromboembolism and mechanical heart valves: a randomized study revisited.

Abstract

This study was designed to revise and substantiate previous inferences, based on short-term follow-up, about differences in the incidence of anticoagulant-related events after heart valve replacement among patients who had been randomly assigned to receive either a Björk-Shiley, Edwards-Duromedics, or Medtronic-Hall mechanical heart valve prosthesis. Intermediate-term follow-up to January 1995 was completed in 418 of 419 patients randomized to receive one of three types of heart valve prostheses between January 1982 and January 1987. Median follow-up was 98.5 months. Multivariable analysis in the hazard function domain was performed to identify factors that influenced the incidence of time-related thromboembolism and bleeding. These findings were compared with those made previously after a median follow-up of 37.5 months. No differences were found among the three prostheses in rates of anticoagulant-related hemorrhage. However, the incidence of thromboembolism was higher after mitral valve replacement among patients who had received the Medtronic-Hall prosthesis (linearized rate, 5.4% per patient year; 70% confidence interval, 4.0% to 7.1%), compared with Edwards-Duromedics (1.3%; 70% confidence interval, 0.4% to 3.0%) and Björk-Shiley prostheses (1.2%; 70% confidence interval, 0.6% to 2.2%). At long-term follow-up, in contrast to the findings at short-term follow-up, patients with either Björk-Shiley or Edwards-Duromedics prostheses had low rates of thromboembolism, whereas higher rates occurred in patients with a Medtronic-Hall prosthesis in the mitral position.