Pharmacovigilance data mining with methods based on false discovery rates: a comparative simulation study.

Area of Science:

• Pharmacovigilance and Drug Safety
• Biostatistics
• Postmarketing Surveillance

Background:

• Postmarketing surveillance is crucial for detecting adverse drug reactions (ADRs) from marketed drugs.
• Current signal detection methods in pharmacovigilance have limitations, including arbitrary detection thresholds.
• False discovery rate (FDR) estimation offers a promising alternative for more robust signal detection.

Purpose of the Study:

• To compare the false-positive performance of existing pharmacovigilance signal detection methods with novel FDR-based approaches.
• To evaluate the effectiveness of FDR-based methods in reducing false positives and signal variability.

Main Methods:

• Two simulation procedures were employed to assess signal detection methods.
• Compared traditional methods (Reporting Odds Ratio, Information Component, Gamma Poisson Shrinkage) against two FDR-based methods (GPS-FDR, Fisher's test-FDR).
• Analyzed false discovery rate (FDR) performance, particularly for signals with at least three reports.

Main Results:

• Significant variability in FDR rates was observed among currently used signal detection methods (0.01% to 12%).
• FDR-based approaches demonstrated a reduction in FDR when analyzing fixed-size lists of signals.
• The number of signals generated by different methods showed considerable variability.

Conclusions:

• FDR-based signal detection methods offer improved effectiveness over traditional approaches in pharmacovigilance.
• The FDR-based Gamma Poisson Shrinkage (GPS) method is particularly promising for enhancing postmarketing surveillance.
• These findings suggest a more reliable way to identify potential drug safety concerns.