Abstract
We assess stratum (e.g. treatment) interactions with covariates and with the baseline hazard function in the proportional hazards (PH) regression model for lifetime data. We consider models incorporating stratum interactions both with and without stratification of the risk sets in the likelihood function, and describe likelihood ratio statistics for tests of the presence of these interactions. We also present step‐down methods for building reduced models which include stratum‐specific parameters corresponding to covariates which interact with treatment. We apply PH models with such interactions to a clinical trial of DES in the treatment of prostate cancer to determine optimal treatment conditional on each patient's covariates.
Original language | English (US) |
---|---|
Pages (from-to) | 73-83 |
Number of pages | 11 |
Journal | Statistics in Medicine |
Volume | 5 |
Issue number | 1 |
DOIs | |
State | Published - 1986 |
Externally published | Yes |
Keywords
- Interaction
- Optimal treatment selection
- Proportional hazards regression model
- Step‐down model selection
- Stratification
- Survival data
ASJC Scopus subject areas
- Epidemiology
- Statistics and Probability