Death and other time-related events after valve replacement.

Abstract

A total of 1533 patients received primary aortic and/or mitral valve replacement, with or without tricuspid valve surgery or other associated cardiac procedures, from January 1, 1975, to July 1, 1979. Actuarial survival at 5 years was 74%. The hazard function (instantaneous risk) of death was highest immediately after operation. Among the incremental risk factors for premature death from any cause were preoperative NYHA class, any valve lesion other than aortic stenosis, aortic cross-clamp time, and combined aortic and mitral valve replacement. Ten modes of death were identified, each with its unique hazard function and risk factors. The commonest mode was cardiac failure. Sudden death was the next commonest, and the early phase of its hazard function peaked about 3 weeks after operation. Seventy-five patients (6.7% of the 1533) had 103 valve reoperations in the follow-up period. Actuarial survival after reoperation was less than that after the primary operation; whether the reoperation was the first, second, or third was a risk factor. Actuarial freedom from prosthetic valve endocarditis (PVE) was 97% at 1 year, and that of periprosthetic leakage without evident infection was 98.8%; the hazard function for the event PVE had an early phase peaking at about 6 weeks and a constant low phase throughout, whereas that of periprosthetic leakage had a single early peaking phase. After reoperations, the actuarial freedom from PVE and periprosthetic leakage was less than that after the original operation. The risk factors for the development of periprosthetic leakage without evident infection were similar but not identical to those for PVE. Among patients receiving a bioprosthesis, 91% were free of bioprosthetic degeneration 5 years after operation. The hazard function was single and continuously rising, and young age and female gender were risk factors. Actuarial freedom from acute prosthetic thrombosis was 97.9% at 5 years. The single hazard function for this event peaked at 10 months. The risk of thrombosis was higher in female patients. The uses of this integrated study, and the inferences from it, are described.