
217Formulating the Required Equations
of time-to-event response variables. This last feature is what distinguishes
this technique from more broadly used techniques like Kaplan–Meier analy-
ses and Cox regression. The Kaplan–Meier analysis produces an empirical
estimate of the distribution of event times without making any assumption
about the underlying shape of the risk function and, hence, is said to be non-
parametric. Cox regression, which relates the hazard of the event over time,
h(t) to various determinants (X
j
), assumes only that the effects of these are
multiplicative on the hazard and remain constant over time. It is written as
follows: ...