Statistical analysis of the results of heart valve replacement.

Abstract

A variety of statistical methods can be used to analyze the results of heart valve replacement. In this review, we illustrate the methodology and the application of the techniques that are most widely used. For early events, univariate analysis and multivariate logistic regression are illustrated. For late (time-related) events, nonparametric methods (Kaplan-Meier and cumulative incidence or 'actual' analysis), parametric methods (based on the exponential, Gompertz and Weibull distributions) and semiparametric methods (Cox proportional hazards) are illustrated.