Related Experiment Videos
A decision-theoretic evaluation of early stopping rules.
D F Heitjan1, P S Houts, H A Harvey
1Center for Biostatistics and Epidemiology, Pennsylvania State University College of Medicine, Hershey 17033.
Statistics in Medicine
|March 1, 1992
Summary
Decision theory offers a framework for optimizing early stopping rules in clinical trials. This approach can lead to significant changes in statistical criteria for trial termination, improving research outcomes.
Area of Science:
- Biostatistics
- Clinical Trial Design
- Decision Theory
Background:
- Early stopping rules in clinical trials are crucial for ethical and efficient research.
- Standard group-sequential tests may not align with all research objectives.
Purpose of the Study:
- To evaluate early stopping rules for clinical trials using a decision-theoretic approach.
- To develop methods for constructing optimal group-sequential tests based on specific utility functions and prior distributions.
Main Methods:
- Modeling a hypothetical phase III, two-arm clinical trial with interim analysis as a decision problem.
- Developing methods to find optimal tests for given utilities/priors and vice versa.
- Constructing optimal tests for diverse perspectives (e.g., maximizing response rate, maximizing correct decisions, expert opinion).
Main Results:
- Optimal early stopping rules are sensitive to input utilities and priors.
- Characteristics of optimal rules can differ significantly from standard group-sequential tests.
- Standard tests may impose an inappropriate symmetry inconsistent with response-rate maximization.
Conclusions:
- Decision theory provides valuable insights into balancing conflicting goals in clinical research.
- Applying decision theory to clinical trial design can lead to substantial modifications in statistical criteria for early termination.