Research Article

Overall C as a measure of discrimination in survival analysis: model specific population value and confidence interval estimation

  1. Michael J. Pencina*,
  2. Ralph B. D'Agostino

Article first published online: 14 JUN 2004

DOI: 10.1002/sim.1802

Issue

Statistics in Medicine

Statistics in Medicine

Volume 23, Issue 13, pages 2109–2123, 15 July 2004

Additional Information

How to Cite

Pencina, M. J. and D'Agostino, R. B. (2004), Overall C as a measure of discrimination in survival analysis: model specific population value and confidence interval estimation. Statistics in Medicine, 23: 2109–2123. doi: 10.1002/sim.1802

Author Information

  1. Statistics and Consulting Unit, Department of Mathematics and Statistics, Boston University, 111 Cummington St. Boston, MA 02215, U.S.A.

*Statistics and Consulting Unit, Department of Mathematics and Statistics, Boston University, 111 Cummington St. Boston, MA 02215, U.S.A.

Publication History

  1. Issue published online: 14 JUN 2004
  2. Article first published online: 14 JUN 2004
  3. Manuscript Accepted:
  4. Manuscript Received:

Funded by

  • NHLBI/NIH. Grant Number: N01-HC-25195

SEARCH

SEARCH BY CITATION

ARTICLE TOOLS

Get PDF (115K)

Keywords:

  • discrimination;
  • performance;
  • survival;
  • ROC;
  • Kendall's τ;
  • concordance

Abstract

The assessment of the discrimination ability of a survival analysis model is a problem of considerable theoretical interest and important practical applications. This issue is, however, more complex than evaluating the performance of a linear or logistic regression. Several different measures have been proposed in the biostatistical literature. In this paper we investigate the properties of the overall C index introduced by Harrell as a natural extension of the ROC curve area to survival analysis. We develop the overall C index as a parameter describing the performance of a given model applied to the population under consideration and discuss the statistic used as its sample estimate. We discover a relationship between the overall C and the modified Kendall's τand construct a confidence interval for our measure based on the asymptotic normality of its estimate. Then we investigate via simulations the length and coverage probability of this interval. Finally, we present a real life example evaluating the performance of a Framingham Heart Study model. Copyright © 2004 John Wiley & Sons, Ltd.

Get PDF (115K)