Biased Estimates of Effect when a Strong but Balanced Covariate is Omitted from a Proportional Hazards Model

@inproceedings{Ishak2007BiasedEO,
  title={Biased Estimates of Effect when a Strong but Balanced Covariate is Omitted from a Proportional Hazards Model},
  author={K. M. A. K. Ishak and James A. Hanley and J. Jaime Caro},
  year={2007}
}
In the analysis of epidemiologic data, potential confounders that are balanced between study groups are often omitted from models estimating the effect of the determinant of interest. An extreme example is that of randomized trials where all potential confounders are omitted if groups appear balanced at baseline due to the randomization of patients. It has… CONTINUE READING