Related Experiment Videos
A SAS macro for sample size re-estimation.
D Zellner1, G E Zellner, F Keller
1Division of Nephrology, Medical Department, University of Ulm, Ulm, Germany. dzellner@t-online.de
Computer Methods and Programs in Biomedicine
|May 8, 2001
Summary
Clinical trial sample size calculations need variance estimates. An internal pilot study using early patient data can help estimate variance and adjust sample size without unblinding treatment status.
Area of Science:
- Clinical Trials
- Biostatistics
- Statistical Methods
Background:
- Accurate sample size determination is crucial for clinical trial power.
- Estimating variance for the primary efficacy variable is often challenging at trial initiation.
- Lack of reliable variance data can lead to underpowered or overly large trials.
Purpose of the Study:
- To present a method for estimating sample size in clinical trials when initial variance estimates are unavailable.
- To introduce an internal pilot study approach for variance estimation.
- To provide a SAS macro for interim power evaluations without unblinding.
Main Methods:
- Utilizing data from the initial patients (internal pilot) to estimate variance.
- Employing the Expectation-Maximization (EM) algorithm for variance estimation.
- Developing a SAS macro to perform simulations and power evaluations.
Main Results:
- The proposed internal pilot method allows for sample size recalculation.
- The EM algorithm facilitates variance estimation without compromising treatment group blinding.
- Simulations demonstrate the utility of the SAS macro for interim power assessments.
Conclusions:
- Internal pilot studies offer a viable solution for sample size adjustments in clinical trials.
- The presented SAS macro provides a practical tool for statistical power management.
- This approach enhances the efficiency and reliability of clinical trial design.