P205: Evaluating MedDRA Query Approaches for Anaphylaxis Case Retrieval in the FDA Adverse Event Monitoring System (AEMS)
Poster Presenter
Anran Liu
ORISE Postdoctoral Fellow
FDA United States
Objectives
To evaluate and compare the case retrieval performance of narrow, broad, and customized MedDRA search strategies for anaphylaxis detection in AEMS.
Method
Using four anaphylaxis MedDRA-based search strategies, we retrieved 2,826 adverse events reports with 27 FDA-approved products. Two independent reviewers assessed each report for confirmed anaphylaxis and product causality. Performance metrics were calculated and compared by search strategy.
Results
The MedDRA-based search strategies included: 1) Preferred Term (PT) Anaphylactic reaction, 2) narrow Anaphylactic reaction Standardised MedDRA Query (SMQ), 3) algorithmic Anaphylactic reaction SMQ, or 4) Anaphylactic reaction Office of New Drugs Custom Medical Query (OCMQ). The search strategies showed nested retrieval patterns: PT Anaphylactic reaction
Conclusion
This study demonstrated varied performance across four MedDRA-based search strategies for identifying anaphylaxis cases in AEMS.
The results reveal a trade-off between sensitivity and specificity across search strategies. The broadest algorithmic SMQ search captured the largest number of true cases, nearly double that of other strategies; but this comprehensive capture came at a substantial cost, requiring review of the most reports with the lowest PPV (25.3%) and highest false positive rate. Conversely, the PT search demonstrated the highest PPV (49.5%) and specificity (85.2%) but lowest sensitivity (43.0%), missing over half of true confirmed anaphylaxis cases retrieved by the algorithmic SMQ.
The OCMQ emerged as strategy that offered the best balance of sensitivity, specificity, and PPV. It identified nearly the same number of true cases as the narrow SMQ (375 vs. 378) with comparable sensitivity (52.5% vs. 52.9%) but achieved substantially higher specificity (80.7% vs. 71.0%) and PPV (48.0% vs. 38.1%). This translated to a meaningful reduction in the number of reports requiring manual review (209 fewer reports than the narrow SMQ) without sacrificing case detection capability.
These findings have important practical implications for pharmacovigilance activities. Pharmacovigilance professionals can select their search strategy based on specific review objectives and available resources. The algorithmic SMQ is appropriate when comprehensive case capture is paramount and resources permit extensive manual review. For routine safety surveillance requiring efficiency, the OCMQ provides an optimal balance between case detection and review burden. The PT is suitable for focused queries prioritizing precision over sensitivity.
Disclaimer: This abstract reflects the views of the authors and should not be construed to represent FDA’s views or policies.
