strcs: A command for fitting flexible parametric survival models on the log-hazard scale

• Bower, Hannah ^[1] ; Crowther, Michael J. ^[1] ; Lambert, Paul C. ^[1]
  1. [1] Department of Medical Epidemiology and Biostatistics Karolinska Institutet
• Localización: The Stata journal, ISSN 1536-867X, Vol. 16, Nº. 4, 2016, págs. 989-1012
• Idioma: inglés
• DOI: 10.1177/1536867x1601600410
• Enlaces
  • Texto completo
• Resumen

  • In this article, we describe strcs, a user-written command for fitting flexible parametric survival models on the log-hazard scale. strcs is an extension of the user-written stgenreg command (Crowther and Lambert, 2013b, Journal of Statistical Software 53(12): 1–17), which fits general parametric models with user-defined hazard functions using numerical integration. strcs implements a two-step method that incorporates both analytical and numerical integration to estimate the cumulative hazard function required for the log-likelihood function. This method improves the accuracy of the fully numeric estimation implemented in stgenreg. Time-dependent effects can be incorporated, and excess mortality models can be fit by using the available options. We also describe some of the extensive postestimation commands that are easily implemented after fitting an strcs model.