|
|||||
|
|
Effect of censoring on adjusting for covariates in comparison of survival times |
|
| Authors: | Timothy.M. Morgan Robert.M Elasoff |
| Affiliation: | 1. Oncology Research Center, Bowman Gray School of Medicine , Wake Forest University , Winston-Salem, NC, 27103;2. Department of Biomathematics , University of California , Los Angeles, CA, 90024 |
| Abstract: | The increase in the variance of the estimate of treatment effect which results from omitting a dichotomous or continuous covariate is quantified as a function of censoring. The efficiency of not adjusting for a covariate is measured by the ratio of the variance obtained with and without adjustment for the covariate. The variance is derived using the Weibull proportional hazards model. Under random censoring, the efficiency of not adjusting for a continuous covariate is an increasing function of the percentage of censored observations. |
| Keywords: | Weibull proportional hazards model clinical trials precision of maximum likelihood estimator failure time data |