RPHVECTORS procedure

Forms vectors for fitting a proportional hazards model as a generalized linear model (R.W. Payne).


Options

SUBJECTS = factor
Subject corresponding to each observation

TIMES = factor or variate
Time of each observation

CENSORED = variate
Contains the value 1 for censored observations, otherwise 0; if unset it is assumed that there is no censoring

RESPONSE = variate
Response variate for the generalized linear model

OFFSET = variate
Offset variate

INDEX = variate
Mapping variate used to produce the expanded variables

NEWSUBJECTS = factor
Expanded subjects factor

NEWTIMES = factor or variate
Expanded times factor

NEWOFFSET = variate
Offset variate for fitting the proportional hazards model


Parameters

X = variates or factors
Lists the x-variables that are to be expanded

NEWX = variates or factors
Identifiers to store the expanded x-variables; if no NEWX is specified, the expanded values overwrite the original values of X


Description

The data for a proportional hazards model consist of a set of subjects observed at one or more times. The final time is usually at the time of death (or failure), otherwise (if the subject survives the trial) the observation is said to be censored. The CENSORED option of RPHVECTORS can be used to specify a variate with an entry for each subject containing one when there is censoring, and zero when there is no censoring. If this is not specified, it is assumed that there is no censoring. The SUBJECTS option can specify a factor to indicate the subject corresponding to each observation; this can be omitted if there is only one observation per subject. The time at which each observation was made must be specified by the TIMES option, in either a factor or a variate.

   The proportional hazards model (Cox 1972) can be fitted in GenStat using the RPHFIT procedure. To fit this as a generalized linear model, the vectors in the model must be expanded so that, for each subject, there is a unit for every time interval up to the last one during which the subject was observed. If (as usually happens) the subject was not observed at every cutpoint, the covariates and treatment factors are taken to be constant during the intervals between the times of the observations.

   RPHVECTORS is used by RPHFIT to produce these expanded vectors, and is made available as a Library procedure in its own right to allow you to program your own proportional hazards analyses. The variates and factors to be expanded are specified using the X parameter, and the NEWX parameter can specify identifiers for the each expanded variate or factor. If NEWX is not specified, the expanded version will replace the original one. The INDEX option allows the variate of indexes used to produce the expanded vectors from the original ones to be saved. The NEWTIMES option can save an expanded factor indicating the time interval corresponding to each unit of the expanded vectors, and the NEWSUBJECTS option can save an expanded factor indicating the subject corresponding to each of their units. The RESPONSE option saves a variate containing one or zero according to whether or not the relevant subject "responded" (i.e. died or failed) during the time period corresponding to each unit of the expanded vectors.

   The generalized linear model also needs an offset variate, containing the log of the exposure time corresponding to each units of the expanded variables. This can be saved using the NEWOFFSET option. This will incorporate any additional offset, which can be specified by the OFFSET option.

 

Options: SUBJECTS, TIMES, CENSORED, RESPONSE, OFFSET, INDEX, NEWSUBJECTS, NEWTIMES, NEWOFFSET.

Parameters: X, NEWX.


Method

RPHVECTORS uses the standard GenStat manipulation commands.


Action with RESTRICT

None of the vectors must be restricted, and any restrictions will be cancelled.


Reference

Cox, D.R. (1972). Regression models and life tables (with discussion). Journal of the Royal Statistical Society B, 34, 187-220.