Bayesian enrollment modeling for several emergency medicine clinical trials

Area of Science:

• Clinical trial management
• Statistical modeling
• Public health research

Background:

• Participant accrual is crucial for clinical trial progress and feasibility.
• Monitoring enrollment provides insights into study timelines and sample size achievement.
• Understanding accrual patterns aids in managing key study events like interim analyses.

Purpose of the Study:

• To develop a flexible statistical framework for characterizing temporal enrollment rates in clinical trials.
• To provide a standardized method for assessing trial progress based on historical accrual data.
• To create a model that can adapt to variable accrual patterns without overreacting to expected or unexpected fluctuations.

Main Methods:

• Utilized a Bayesian first-order simple dynamic linear model.
• Employed weakly informative priors for model estimation.
• Characterized enrollment rates temporally within pre-defined quarterly time windows.

Main Results:

• The model successfully characterized accrual patterns in three ongoing clinical trials.
• Demonstrated flexibility in adapting to variable enrollment rates.
• Showed robustness against expected (e.g., seasonal) and unexpected (e.g., pandemic) accrual variability.

Conclusions:

• The presented model offers a valuable addition to statistical literature on clinical trial analysis.
• Provides a flexible and standardized framework for characterizing trial accrual trends.
• This approach is beneficial for study teams, sponsors, and funding agencies interested in assessing trial progress.