Abstract

The maximized Sequential Probability Ratio Test (MaxSPRT)
is broadly utilized for monitoring adverse events following
post-market drugs and vaccines. The basic assumption is that
the increased relative risk of adverse events, if any, is
constant among the individuals of different strata of
categorical covariates, such as gender and age. However,
it may be the case that the different strata present unequal
relative risks. The present work introduces an adjusted MaxSPRT
for the monitoring and detection of heterogeneous relative risk
structures among subgroups of individuals formed by the combination
of categorical covariates. The signaling threshold is exact and
obtained according to arbitrary alpha spending functions.