Flexible Parametric Modelling of the Hazard Function in Breast Cancer Studies

In cancer research, study of the hazard function provides useful information on the disease dynamic, in addition to the identification of prognostic factors. The widely utilized Cox proportional hazard model uses a stepwise nonparametric estimator for the baseline hazard function. Therefore the use of parametric models and/or other approaches to estimate the hazard function is often invoked. A recent work by C. Cox and colleagues [1] has stimulated the use of a complex and flexible parametric model based on the General Gamma distribution, supported by the development of optimization software. Use of the General Gamma to study the shape of the hazard function is investigated. As a benchmark, the flexible approach based on piecewise exponential model and a nonparametric kernel estimate are considered. An example based on breast cancer survival is used to illustrate the main findings.