Covariate-adjusted measures of discrimination for survival data

for the Emerging Risk Factors Collaboration

Research output: Contribution to journalArticle

6 Citations (Scopus)

Abstract

Motivation: Discrimination statistics describe the ability of a survival model to assign higher risks to individuals who experience earlier events: examples are Harrell's C-index and Royston and Sauerbrei's D, which we call the D-index. Prognostic covariates whose distributions are controlled by the study design (e.g. age and sex) influence discrimination and can make it difficult to compare model discrimination between studies. Although covariate adjustment is a standard procedure for quantifying disease-risk factor associations, there are no covariate adjustment methods for discrimination statistics in censored survival data. Objective: To develop extensions of the C-index and D-index that describe the prognostic ability of a model adjusted for one or more covariate(s). Method: We define a covariate-adjusted C-index and D-index for censored survival data, propose several estimators, and investigate their performance in simulation studies and in data from a large individual participant data meta-analysis, the Emerging Risk Factors Collaboration. Results: The proposed methods perform well in simulations. In the Emerging Risk Factors Collaboration data, the age-adjusted C-index and D-index were substantially smaller than unadjusted values. The study-specific standard deviation of baseline age was strongly associated with the unadjusted C-index and D-index but not significantly associated with the age-adjusted indices. Conclusions: The proposed estimators improve meta-analysis comparisons, are easy to implement and give a more meaningful clinical interpretation.

Original languageEnglish
Pages (from-to)592-613
Number of pages22
JournalBiometrical Journal
Volume57
Issue number4
DOIs
Publication statusPublished - Jul 1 2015

Fingerprint

Survival Data
Discrimination
Covariates
Risk Factors
Censored Survival Data
Adjustment
Model Discrimination
Statistics
Estimator
Survival Model
Standard deviation
Assign
Baseline
Simulation Study

All Science Journal Classification (ASJC) codes

  • Statistics and Probability
  • Statistics, Probability and Uncertainty

Cite this

Covariate-adjusted measures of discrimination for survival data. / for the Emerging Risk Factors Collaboration.

In: Biometrical Journal, Vol. 57, No. 4, 01.07.2015, p. 592-613.

Research output: Contribution to journalArticle

for the Emerging Risk Factors Collaboration 2015, 'Covariate-adjusted measures of discrimination for survival data', Biometrical Journal, vol. 57, no. 4, pp. 592-613. https://doi.org/10.1002/bimj.201400061
for the Emerging Risk Factors Collaboration. / Covariate-adjusted measures of discrimination for survival data. In: Biometrical Journal. 2015 ; Vol. 57, No. 4. pp. 592-613.
@article{f97da0d02e8e4fdb94e97d6676636cf5,
title = "Covariate-adjusted measures of discrimination for survival data",
abstract = "Motivation: Discrimination statistics describe the ability of a survival model to assign higher risks to individuals who experience earlier events: examples are Harrell's C-index and Royston and Sauerbrei's D, which we call the D-index. Prognostic covariates whose distributions are controlled by the study design (e.g. age and sex) influence discrimination and can make it difficult to compare model discrimination between studies. Although covariate adjustment is a standard procedure for quantifying disease-risk factor associations, there are no covariate adjustment methods for discrimination statistics in censored survival data. Objective: To develop extensions of the C-index and D-index that describe the prognostic ability of a model adjusted for one or more covariate(s). Method: We define a covariate-adjusted C-index and D-index for censored survival data, propose several estimators, and investigate their performance in simulation studies and in data from a large individual participant data meta-analysis, the Emerging Risk Factors Collaboration. Results: The proposed methods perform well in simulations. In the Emerging Risk Factors Collaboration data, the age-adjusted C-index and D-index were substantially smaller than unadjusted values. The study-specific standard deviation of baseline age was strongly associated with the unadjusted C-index and D-index but not significantly associated with the age-adjusted indices. Conclusions: The proposed estimators improve meta-analysis comparisons, are easy to implement and give a more meaningful clinical interpretation.",
author = "{for the Emerging Risk Factors Collaboration} and White, {Ian R.} and Eleni Rapsomaniki and Wannamethee, {S. G.} and Morris, {R. W.} and J. Willeit and P. Willeit and P. Santer and S. Kiechl and N. Wald and S. Ebrahim and Lawlor, {D. A.} and J. Gallacher and Yarnell, {J. W.G.} and Y. Ben-Shlomo and E. Casiglia and V. Tikhonoff and Sutherland, {S. E.} and Nietert, {P. J.} and Keil, {J. E.} and Bachman, {D. L.} and Psaty, {B. M.} and M. Cushman and Nordestgaard, {B. G.} and A. Tybj{\ae}rg-Hansen and R. Frikke-Schmidt and S. Giampaoli and L. Palmieri and S. Panico and L. Pilotto and D. Vanuzzo and Simons, {L. A.} and Y. Friedlander and J. McCallum and Price, {J. F.} and S. McLachlan and Taylor, {J. O.} and Guralnik, {J. M.} and Wallace, {R. B.} and Kohout, {F. J.} and Cornoni-Huntley, {J. C.} and Guralnik, {J. M.} and Blazer, {D. G.} and Guralnik, {J. M.} and Phillips, {C. L.} and Phillips, {C. L.} and Guralnik, {J. M.} and Wareham, {N. J.} and Khaw, {K. T.} and H. Brenner and B. Sch{\"o}ttker",
year = "2015",
month = "7",
day = "1",
doi = "10.1002/bimj.201400061",
language = "English",
volume = "57",
pages = "592--613",
journal = "Biometrical Journal",
issn = "0323-3847",
publisher = "Wiley-VCH Verlag",
number = "4",

}

TY - JOUR

T1 - Covariate-adjusted measures of discrimination for survival data

AU - for the Emerging Risk Factors Collaboration

AU - White, Ian R.

AU - Rapsomaniki, Eleni

AU - Wannamethee, S. G.

AU - Morris, R. W.

AU - Willeit, J.

AU - Willeit, P.

AU - Santer, P.

AU - Kiechl, S.

AU - Wald, N.

AU - Ebrahim, S.

AU - Lawlor, D. A.

AU - Gallacher, J.

AU - Yarnell, J. W.G.

AU - Ben-Shlomo, Y.

AU - Casiglia, E.

AU - Tikhonoff, V.

AU - Sutherland, S. E.

AU - Nietert, P. J.

AU - Keil, J. E.

AU - Bachman, D. L.

AU - Psaty, B. M.

AU - Cushman, M.

AU - Nordestgaard, B. G.

AU - Tybjærg-Hansen, A.

AU - Frikke-Schmidt, R.

AU - Giampaoli, S.

AU - Palmieri, L.

AU - Panico, S.

AU - Pilotto, L.

AU - Vanuzzo, D.

AU - Simons, L. A.

AU - Friedlander, Y.

AU - McCallum, J.

AU - Price, J. F.

AU - McLachlan, S.

AU - Taylor, J. O.

AU - Guralnik, J. M.

AU - Wallace, R. B.

AU - Kohout, F. J.

AU - Cornoni-Huntley, J. C.

AU - Guralnik, J. M.

AU - Blazer, D. G.

AU - Guralnik, J. M.

AU - Phillips, C. L.

AU - Phillips, C. L.

AU - Guralnik, J. M.

AU - Wareham, N. J.

AU - Khaw, K. T.

AU - Brenner, H.

AU - Schöttker, B.

PY - 2015/7/1

Y1 - 2015/7/1

N2 - Motivation: Discrimination statistics describe the ability of a survival model to assign higher risks to individuals who experience earlier events: examples are Harrell's C-index and Royston and Sauerbrei's D, which we call the D-index. Prognostic covariates whose distributions are controlled by the study design (e.g. age and sex) influence discrimination and can make it difficult to compare model discrimination between studies. Although covariate adjustment is a standard procedure for quantifying disease-risk factor associations, there are no covariate adjustment methods for discrimination statistics in censored survival data. Objective: To develop extensions of the C-index and D-index that describe the prognostic ability of a model adjusted for one or more covariate(s). Method: We define a covariate-adjusted C-index and D-index for censored survival data, propose several estimators, and investigate their performance in simulation studies and in data from a large individual participant data meta-analysis, the Emerging Risk Factors Collaboration. Results: The proposed methods perform well in simulations. In the Emerging Risk Factors Collaboration data, the age-adjusted C-index and D-index were substantially smaller than unadjusted values. The study-specific standard deviation of baseline age was strongly associated with the unadjusted C-index and D-index but not significantly associated with the age-adjusted indices. Conclusions: The proposed estimators improve meta-analysis comparisons, are easy to implement and give a more meaningful clinical interpretation.

AB - Motivation: Discrimination statistics describe the ability of a survival model to assign higher risks to individuals who experience earlier events: examples are Harrell's C-index and Royston and Sauerbrei's D, which we call the D-index. Prognostic covariates whose distributions are controlled by the study design (e.g. age and sex) influence discrimination and can make it difficult to compare model discrimination between studies. Although covariate adjustment is a standard procedure for quantifying disease-risk factor associations, there are no covariate adjustment methods for discrimination statistics in censored survival data. Objective: To develop extensions of the C-index and D-index that describe the prognostic ability of a model adjusted for one or more covariate(s). Method: We define a covariate-adjusted C-index and D-index for censored survival data, propose several estimators, and investigate their performance in simulation studies and in data from a large individual participant data meta-analysis, the Emerging Risk Factors Collaboration. Results: The proposed methods perform well in simulations. In the Emerging Risk Factors Collaboration data, the age-adjusted C-index and D-index were substantially smaller than unadjusted values. The study-specific standard deviation of baseline age was strongly associated with the unadjusted C-index and D-index but not significantly associated with the age-adjusted indices. Conclusions: The proposed estimators improve meta-analysis comparisons, are easy to implement and give a more meaningful clinical interpretation.

UR - http://www.scopus.com/inward/record.url?scp=84933514328&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84933514328&partnerID=8YFLogxK

U2 - 10.1002/bimj.201400061

DO - 10.1002/bimj.201400061

M3 - Article

C2 - 25530064

AN - SCOPUS:84933514328

VL - 57

SP - 592

EP - 613

JO - Biometrical Journal

JF - Biometrical Journal

SN - 0323-3847

IS - 4

ER -