Adaptive Sequential Multiple Hypotheses Testing for Concomitant Vaccine Safety Surveillance.

Area of Science:

• Biostatistics
• Epidemiology
• Vaccinology

Background:

• Simultaneous vaccination recommendations necessitate robust statistical methods for monitoring adverse events.
• Current methods often analyze vaccines individually, potentially missing risks associated with combined exposures.

Purpose of the Study:

• To introduce an adaptive multiple hypotheses testing method for detecting adverse event risks from simultaneous vaccine combinations.
• To extend the binomial maximized sequential probability ratio test (MaxSPRT) to a multinomial model for this purpose.

Main Methods:

• Extension of the binomial MaxSPRT to a multinomial probability model.
• Utilized an exact analytical alpha spending approach for statistical rigor.
• Employed Monte Carlo methods for complex scenarios involving three or more vaccines.

Main Results:

• The method can rapidly detect increased risks of adverse events from single or multiple vaccine exposures.
• An analytical approach is computationally feasible for up to two simultaneous vaccine exposures.
• Monte Carlo simulations validate the approach for more complex vaccination schedules.

Conclusions:

• The developed statistical method offers a more comprehensive approach to vaccine safety monitoring.
• This adaptive testing framework enhances the ability to identify adverse event signals from combined vaccine use.
• The methodology is applicable to real-world public health surveillance of vaccine safety.