A conditional sequential sampling procedure for drug safety surveillance

Area of Science:

• Pharmacovigilance and Drug Safety
• Biostatistics
• Epidemiology

Background:

• Prospective drug safety surveillance relies on periodic data updates.
• Identifying elevated adverse event risks requires robust statistical methods.
• Existing methods may not adequately adjust for confounding factors and temporal trends.

Purpose of the Study:

• To introduce a practical group sequential method for testing drug-induced adverse event risks.
• To provide a flexible approach for ongoing drug safety surveillance using administrative claims data.
• To adjust for multiple testing, temporal trends, and population heterogeneity.

Main Methods:

• A conditional sequential sampling procedure is proposed.
• The method utilizes summary tables of exposed person-times and adverse events, stratified by confounders.
• It incorporates alpha-spending functions and conditions on event counts within strata and time periods.

Main Results:

• Simulations demonstrate the method's performance across various scenarios.
• The approach effectively adjusts for multiple testing and confounding factors.
• Application to Rofecoxib, Diclofenac, and Naproxen data illustrates its utility in real-world safety assessments.

Conclusions:

• The proposed group sequential method offers a flexible and practical solution for drug safety surveillance.
• It reliably detects elevated relative risks while controlling Type I error.
• The method is applicable to diverse settings using periodically updated administrative claims data.