[ SURVIVAL ] [ OPTIONS ] [ OUTPUT_GROUPS ] [ STATEMENTS ]
SURVIVAL provides proportional hazards modeling for failure time outcomes, which may contain left- and right- censored observations, time-dependent covariates, and multiple events per subject. The SURVIVAL procedure fits the discrete or continuous (Cox) proportional hazards model to sample surveys and other clustered data applications. Estimates of the model parameters and their standard errors are computed, along with tests of hypotheses.
SURVIVAL capabilities includes::
Counting process style of input (Andersen and Gill, 1982) to permit left truncation, multiple events per subject, and time-dependent covariates. A time-dependent covariate is one whose value for any given individual can change over time during the course of a study
Computation of Schoenfeld residuals and Score residuals to allow users to evaluate “goodness of fit” and the validity of the proportional hazards assumption
Computation of Efron's likelihood approximation for ties in addition to the current default formula by Breslow
Option to allow stratified baseline hazard functions for different types of failures or different subgroups of the population.
Procedure Enhancements:
The BRR, Delete-1 jackknife and Replicate weight jackknife designs have been implemented for estimating variances of betas and marginals.
Estimates of confidence limits for the model parameters are now produced by default in the BETAS group.
The PREDMARG, CONDMARG, COND_EFF and PRED_EFF statements are now available for estimating predicted and conditional marginals and their variances for effects and interactions.
SURVIVAL computes martingale residuals to allow users to evaluate “goodness of fit” and the validity of the proportional hazards assumption.
SURVIVAL also computes normalized Schoenfeld and score residuals. The normalized score residuals are the taylorized deviations for the betas and may be useful for further analysis.
The CLASS statement is available in SURVIVAL.