Related Experiment Videos
Floating absolute risk: an alternative to relative risk in survival and case-control analysis avoiding an arbitrary
D F Easton1, J Peto, A G Babiker
1Section of Epidemiology, Institute of Cancer Research, Sutton, Surrey, U.K.
Statistics in Medicine
|July 1, 1991
Summary
This study introduces a new method for comparing multiple groups in survival analysis and matched case-control studies. It provides more accurate standard errors, enabling efficient and independent group comparisons without a baseline.
Area of Science:
- Biostatistics
- Epidemiology
- Statistical Modeling
Background:
- Standard methods for multiple group comparisons in survival analysis (Cox model) and matched case-control studies often rely on a baseline group.
- This approach leads to unsatisfactory standard errors and confidence limits due to correlated parameters dependent on the baseline group's precision.
Purpose of the Study:
- To develop a method for describing multiple group comparisons that avoids the limitations of a baseline group.
- To enable efficient and independent comparison of all groups in survival analysis and matched case-control studies.
Main Methods:
- The study describes the construction of standard errors for parameters of all groups, eliminating the need for a designated baseline group.
- These newly constructed standard errors allow for comparisons of groups as if they were roughly independent parameters.
Main Results:
- The proposed method yields standard errors that can be treated as relating to independent parameters.
- This facilitates efficient group comparisons without requiring knowledge of covariances between parameters.
Conclusions:
- The novel method improves the description of multiple group comparisons in survival analysis and matched case-control studies.
- It enhances graphical presentations of relative risks and aids in combining results from multiple published studies.