Adjusting for observable selection bias in block randomized trials.

Area of Science:

• Clinical Trials Methodology
• Biostatistics
• Health Services Research

Background:

• Selection bias can compromise the integrity of randomized clinical trials.
• Observable selection bias occurs when patient enrollment is influenced by factors related to treatment allocation.
• Accurate assessment of treatment effects requires addressing potential biases.

Purpose of the Study:

• To propose and evaluate a model-based approach for detecting and adjusting observable selection bias in two-treatment randomized clinical trials.
• To enable testing for both the presence of selection bias and true treatment effects.

Main Methods:

• A model-based approach was developed to identify and quantify observable selection bias.
• Simulations were conducted using a randomized block design with biased patient enrollment strategies.
• The model was tested for its ability to detect bias and adjust treatment effect estimates.

Main Results:

• The proposed method successfully detected selection bias under simulated biased enrollment conditions.
• The model demonstrated the capability to adjust for observable selection bias.
• The approach allows for testing treatment effects while accounting for potential bias.

Conclusions:

• The developed model-based approach is effective for detecting and adjusting observable selection bias in randomized clinical trials.
• This method enhances the reliability of treatment effect estimations in the presence of enrollment bias.
• The approach provides a valuable tool for improving the validity of clinical trial results.