Covariate-adjusted measures of discrimination for survival data.

Pubmed ID: 25530064

Pubmed Central ID: PMC4666552

Journal: Biometrical journal. Biometrische Zeitschrift

Publication Date: July 1, 2015

Affiliation: Farr Institute for Health Informatics Research, Department of Epidemiology and Public Health, University College London Medical School, 222 Euston Road, London WC1E 6BT, UK.

MeSH Terms: Humans, Male, Female, Cardiovascular Diseases, Risk Factors, Middle Aged, Survival Analysis, Clinical Trials as Topic, Analysis of Variance, Biometry, Discriminant Analysis

Grants: RG/08/014/24067, MR/K013351/1, UL1 TR000062, MR/L003120/1, G0700463, MC_UU_12013/5, UL1 TR001450, G19/35, G0100222, G8802774, G0902037, G1000616, RG/07/008/23674, RG/13/13/30194

Authors: Zhang Y, Franco OH, Woodward M, Mathiesen EB, Tuomainen TP, Kiechl S, Hofman A, Kauhanen J, Ducimetiere P, Willeit J, Ballantyne CM, Davis BR, Howard BV, Trevisan M, Coresh J, Gao P, Kaptoge S, Cushman M, Cushman M, Collins R, Danesh J, Khaw KT, Vasan RS, Koenig W, Brenner H, Rothenbacher D, Meisinger C, Barrett-Connor E, Daniels LB, Walker M, Wood A, Knuiman MW, Arima H, Kiyohara Y, Simons LA, Nissinen A, Lappas G, Lappas G, Rosengren A, Hata J, Ninomiya T, Wilhelmsen L, Assmann G, Schulte H, Lissner L, Wilsgaard T, Onat A, Kivimaki M, Geleijnse JM, Feskens EJ, Shimbo D, Pennells L, Lawlor DA, Nordestgaard BG, Salomaa V, Sattar N, Folsom AR, Folsom AR, Atkins R, Barr EL, Shaw JE, Zimmet PZ, Wannamethee SG, Morris RW, Ebrahim S, Gallacher J, Nietert PJ, Sutherland SE, Bachman DL, Keil JE, de Boer IH, Tybjærg-Hansen A, Frikke-Schmidt R, Giampaoli S, Giampaoli S, Palmieri L, Panico S, Vanuzzo D, Pilotto L, Blazer DG, Guralnik JM, Guralnik JM, Guralnik JM, Guralnik JM, Phillips CL, Phillips CL, Harald K, Vartiainen E, Dankner R, Chetrit A, Lubin F, Eriksson H, Svärdsudd K, Welin L, Björkelund C, Dekker JM, Nijpels G, Stehouwer CD, Iso H, Iso H, Kitamura A, Kitamura A, Yamagishi K, Noda H, Psaty BM, Psaty BM, Verschuren WM, Verschuren WM, Blokstra A, Blokstra A, Bueno-de-Mesquita HB, Fletcher A, Davidson KW, Kirkland S, Shaffer JA, Sato S, Bakker SJ, Gansevoort RT, Amouyel P, Arveiler D, Evans A, Ferrières J, Wingard DL, Tunstall-Pedoe H, Tavendale R, Lowe GD, Umans J, Nakagawa H, Sakurai M, Nakamura K, Morikawa Y, Njølstad I, Sundström J, Brunner E, Shipley M, Buring J, Butterworth AS, White IR, Daimon M, Price JF, Salonen JT, Willeit P, Willeit P, McLachlan S, Grandits G, Kuller LH, Peters A, Wareham NJ, Ford I, Ridker PM, Ridker PM, Simpson LM, Cook N, Rapsomaniki E, Tipping RW, Santer P, Wald N, Yarnell JW, Ben-Shlomo Y, Ben-Shlomo Y, Casiglia E, Tikhonoff V, Friedlander Y, McCallum J, Taylor JO, Wallace RB, Kohout FJ, Cornoni-Huntley JC, Schöttker B, Müller HT, Donfrancesco C, Jousilahti PR, D' Agostino RB, Wolf PA, Oizumi T, Kayama T, Kato T, Mehlig K, Nagel D, Rodriguez B, Goldbourt U, DeStavola BL, Cooper JA, Bauer KA, Dullaart RP, Jukema JW, Westendorp RG, Cantin B, Lamarche B, Després JP, Gudnason V, Aspelund T, Howard WJ, Best LG, Davey-Smith G, Gaziano JM, Marmot M, Clarke R, Rifai N, Robertson M, Ibañez Marín A, Bolton T, Burgess S, di Angelantonio E, Harshfield E, Peters S, Spackman S, Thompson S, White I

Cite As: White IR, Rapsomaniki E, Emerging Risk Factors Collaboration. Covariate-adjusted measures of discrimination for survival data. Biom J 2015 Jul;57(4):592-613. Epub 2014 Dec 20.

Studies:

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.