Benefits and limitations of Kaplan-Meier calculations of survival chance in cancer surgery

Abstract

Background and aim: Especially in malign diseases, the therapeutic decision depends on the prognosis for the individual patient. A prognosis is a prediction of the future course of disease following its onset. Graphical representation of such statistical results-such as the well-known Kaplan-Meier curve-is often used to assist readers of a paper in the interpretation. However, mistakes and distortions frequently arise in the display and interpretation of survival plots. This review aims to highlight such pitfalls and provide recommendations for future practice.

Methods: Special topics are discussed: the criteria for the presentation of the survival curve, the problem of missing values, estimation of the prognosis in the presence of competing risks, comparison of treatment effects and analysis of survival by tumour-response category.