Alpha spending for historical versus surveillance Poisson data with CMaxSPRT

Area of Science:

• Pharmacovigilance
• Biostatistics
• Epidemiology

Background:

• Sequential analysis is vital for postmarket drug and vaccine safety surveillance.
• The conditional maximized sequential probability ratio test (CMaxSPRT) is a formal solution for Poisson-distributed adverse events with uncertain baseline rates.
• CMaxSPRT traditionally uses a flat signaling threshold.

Purpose of the Study:

• To demonstrate the application of CMaxSPRT under non-flat signaling thresholds.
• To explore the alpha spending approach in the context of CMaxSPRT.
• To provide guidance on optimizing signaling threshold shapes for improved surveillance efficiency.

Main Methods:

• Adaptation of the CMaxSPRT methodology for non-flat thresholds.
• Integration with the alpha spending approach for flexible threshold design.
• Development of a rule of thumb for threshold shape optimization.

Main Results:

• CMaxSPRT can be effectively performed using non-flat signaling thresholds.
• Non-flat thresholds offer advantages in minimizing expected time to signal and sample size.
• The proposed methods were illustrated with an influenza vaccine safety surveillance example.

Conclusions:

• The enhanced CMaxSPRT framework provides a more flexible and efficient tool for pharmacovigilance.
• Optimizing signaling thresholds can significantly improve the performance of sequential safety surveillance.
• This approach is applicable to various postmarket surveillance scenarios for drugs and vaccines.