Point estimation following a two-stage group sequential trial.

Area of Science:

• Biostatistics
• Clinical Trial Design
• Statistical Inference

Background:

• Group sequential trials allow interim analyses, but repeated testing can introduce bias into parameter estimates.
• Adjusted point estimation procedures are crucial for mitigating bias in clinical trial results.

Purpose of the Study:

• To evaluate and compare the performance of nine distinct point estimators in a two-stage group sequential trial setting.
• To analyze conditional and marginal biases and residual mean square error for each estimator.

Main Methods:

• A common general framework was used to describe nine point estimators for two-stage group sequential trials.
• Performance was contrasted across five example trial settings, focusing on bias and mean square error metrics.

Main Results:

• The uniform minimum variance unbiased estimator, while marginally unbiased, exhibited significant conditional bias and residual mean square error.
• Conditional uniform minimum variance unbiased estimator is suitable for inference on trial progression.
• Two mean adjusted estimators demonstrated superior performance regarding marginal residual mean square error.

Conclusions:

• The optimal point estimator choice depends on the desired balance between conditional bias, marginal bias, and residual mean square error.
• Mean adjusted estimators offer a favorable balance for marginal residual mean square error.
• Conditional maximum likelihood estimates are preferable if only conditional and marginal biases are critical.