Figure 1.

A model for which parameter estimation can be based on an event-time table. We assume that the incidence rate h(...) for some event depends only on three time-varying factors, namely age and time since last pregnancy, suitably categorised, and number of births. Each combination of levels of the three time-varying factors defines a stratum within which the incidence rate is constant, leading to piecewise constant incidence rates. The graphs show the changing values of the time-varying factors ("exposures"), the changing values of the stratified time-varying factors (the fat lines) and the resulting incidence rate during follow-up of one person.