A Flexible Parametric Competing-risks Model Using a Direct Likelihood Approach for the Cause-specific Cumulative Incidence Function. (June 2017)

• Record Type: Journal Article
• Authors: Mozumder, Sarwar Islam; Rutherford, Mark J.; Lambert, Paul C.
• Is Part Of: Stata journal. Volume 17:Number 2(2017)
• Page Start: 462
• Page End: 489
• Publication Date: 2017-06
• Subjects: st0482 -- stpm2cr -- survival analysis -- competing risks -- flexible parametric models -- subdistribution hazard -- cumulative incidence function; Statistics -- Periodicals; Statistics -- Computer programs -- Periodicals; 001.422
• DOI: 10.1177/1536867X1701700212
• Languages: English
• ISSNs: 1536-867X

Abstract

In competing-risks analysis, the cause-specific cumulative incidence function (CIF) is usually obtained in a modeling framework by either 1) transforming on all cause-specific hazards or 2) transforming by using a direct relationship with the subdistribution hazard function. We expand on current competing-risks methodology from within the flexible parametric survival modeling framework and focus on the second approach. This approach models all cause-specific CIFs simultaneously and is more useful for answering prognostic-related questions. We propose the direct flexible parametric survival modeling approach for the cause-specific CIF. This approach models the (log cumulative) baseline hazard without requiring numerical integration, which leads to benefits in computational time. It is also easy to make out-of-sample predictions to estimate more useful measures and incorporate alternative link functions, for example, logit links. To implement these methods, we introduce a new estimation command, stpm2cr, and demonstrate useful predictions from the model through an illustrative melanoma dataset.