ABSTRACT
Chapter 4 gives a less technical introduction to models for risks and other marginal parameters in multi-state models, and examples are given to illustrate how results from analyses can be interpreted. The marginal models fall in two categories using either a plug-in approach or direct modelling. Plug-in uses results from intensity models combined with a description of the relationship between the marginal parameter and the intensities. This includes the Kaplan-Meier and Aalen-Johansen estimators for, respectively, the survival function and the competing risks cumulative incidence. In the second approach, a direct model is set up for the marginal parameter and estimates are obtained by solving suitable estimating equations. This includes the Fine-Gray model for the cumulative incidence and the multiplicative Ghosh-Lin regression model for the expected number of recurrent events. Marginal hazard models for dependent data are also introduced.
