Comparison of Cox Model Methods in A Low-dimensional Setting with Few Events

• Published in: Genomics, proteomics & bioinformatics Vol. 14; no. 4; pp. 235 - 243
• Main Authors: Ojeda, Francisco M., Müller, Christian, Börnigen, Daniela, Trégouët, David-Alexandre, Schillert, Arne, Heinig, Matthias, Zeller, Tanja, Schnabel, Renate B.
• Format: Journal Article
• Language: English
• Published: China Elsevier Ltd 01.08.2016; Department of General and Interventional Cardiology, University Heart Center Hamburg-Eppendorf, 20246 Hamburg, Germany%Sorbonne Universite′s, Universite′ Pierre et Marie Curie Paris 06, Institut National pour la Sante′ et la Recherche Me′dicale INSERM, Unite′ Mixte de Recherche en Sante′ UMR_S 1166, F-75013 Paris, France%Institut fu¨r Medizinische Biometrie und Statistik, Universita¨t zu Lu¨beck, Universita¨tsklinikum Schleswig-Holstein, Campus Lu¨beck, 23562 Lu¨beck, Germany%Institute of Computational Biology, German Research Center for Environmental Health, Helmholtz Zentrum Mu¨nchen, 85764 Neuherberg, Germany; Elsevier; Oxford University Press
• Subjects: Bioinformatics; Biomarkers - blood; Biomarkers - metabolism; Computer Science; Coronary artery disease; Coronary Artery Disease - diagnosis; Coronary Artery Disease - genetics; Cox模型; Events per variable; Genetic Variation; Humans; Original Research; Penalized regression; Prognosis; Proportional Hazards Models; Proportional hazards regression; Prospective Studies; 事件; 估计方法; 低维; 冠状动脉疾病; 生物标志物; 设置; 预测模型; Proportional hazards regression; Events per variable; Penalized regression; Coronary artery disease; Proportional hazards regres-sion
• ISSN: 1672-0229; 2210-3244
• DOI: 10.1016/j.gpb.2016.03.006

Prognostic models based on survival data frequently make use of the Cox proportional hazards model. Developing reliable Cox models with few events relative to the number of predictors can be challenging, even in low-dimensional datasets, with a much larger number of observations than variables. In such a setting we examined the performance of methods used to estimate a Cox model, including （i） full model using all available predictors and estimated by standard techniques, （ii） backward elimination （BE）, （iii） ridge regression, （iv） least absolute shrinkage and selection operator （lasso）, and （v） elastic net. Based on a prospective cohort of patients with manifest coronary artery disease （CAD）, we performed a simulation study to compare the predictive accuracy, calibration, and discrimination of these approaches, Candidate predictors for incident cardiovascular events we used included clinical variables, biomarkers, and a selection of genetic variants associated with CAD. The penalized methods, i.e., ridge, lasso, and elastic net, showed a comparable performance, in terms of predictive accuracy, calibration, and discrimination, and outperformed BE and the full model. Excessive shrinkage was observed in some cases for the penalized methods, mostly on the simulation scenarios having the lowest ratio of a number of events to the number of variables. We conclude that in similar settings, these three penalized methods can be used interchangeably. The full model and backward elimination are not recommended in rare event scenarios.