AMFIT is designed for use in modeling hazard functions for rate data organized as a person-year (rate) table and can also be used to work with count data assumed to have a Poisson distribution. The input data consist of records containing cases, person-years, and covariates. The rates to be modeled are computed as the ratio of cases to person-years for each record in the input data set. To fit a model with AMFIT, it is necessary to specify both the cases and person-year variables. The default names for these variables are cases and pyr, respectively. The commands for explicit specification of these key variables are CASES and PYR. Both cases and person-year variables must be specified to fit a model in AMFIT. AMFIT models can include multiple categorical stratification variables, which are specified with the STRATA command and removed with the NOSTRATA command. (Stratification is discussed later in this chapter.) .
As described above, if the input data set contains variables with the default names, and these are the variables you want to use to describe the rates to be modeled, then explicit specification of the key variables is not necessary. However, if, for example, the person-year variable is called py and the cases variable is breastca, then the following commands can be used to specify these variables as the key variables:
CASES breastca @
PYR py @