Related Experiment Videos
Sample-size calculations for the Cox proportional hazards regression model with nonbinary covariates
1CSPCC, Department of Veterans Affairs Palo Alto Health Care System, Palo Alto, CA, USA. fhsieh@mailsvr.icon.palo-alto.med.va.gov
Controlled Clinical Trials
|January 9, 2001
Summary
This study presents a formula for determining sample size in proportional hazards regression with non-binary predictors. It confirms censored data do not impact statistical power, crucial for accurate survival analysis.
Area of Science:
- Biostatistics
- Survival Analysis
- Clinical Trials
Background:
- Proportional hazards regression is a key tool in survival analysis.
- Determining adequate sample size is critical for study power.
- Existing methods often assume binary covariates.
Purpose of the Study:
- To derive a formula for sample size calculation in proportional hazards models with non-binary covariates.
- To investigate the impact of censored observations on statistical power.
- To provide a variance inflation factor for sample size adjustment with additional covariates.
Main Methods:
- Derivation of a novel sample size formula.
- Monte Carlo simulations to assess statistical power.
- Analysis of censored observations in proportional hazards models.
Main Results:
- The derived formula enables sample size calculation for non-binary covariates.
- Simulations confirmed censored observations do not contribute to test power.
- A variance inflation factor was provided for models with multiple covariates.
Conclusions:
- The new formula enhances sample size planning for complex survival analyses.
- Understanding the role of censored data is vital for efficient study design.
- The findings support robust statistical power in clinical trials.