Developing a signal triage algorithm for Thai national adverse drug reaction database

Abstract

Keywords: adverse drug reaction, signal triage, prioritization, signal detection, multi-criteria decision analysis

Objectives: There was increasing adverse drug reaction (ADR) reports submitted to the Health Product Vigilance Center under the Thai FDA. The Thai Signal Detection Program was developed to identify and filter the potential signals, called signals of disproportionate reporting (SDRs). A large number of SDRs cannot be in-depth assessed by the Signal Detection Advisory Working Group (SDAWG) in time. The prioritized SDRs with concentrated in-depth assessment might help find some true signals. This preliminary study aimed at developing a signal triage algorithm that can prioritize SDRs to assign an in-depth assessment.

Methods: A multi-criteria decision analysis (MCDA) was chosen for a triage algorithm by generating scores for priority rankings on clinical importance of SDRs. This study had three main steps. Key attributes for a triage decision was first identified and followed by the development of a signal triage algorithm. After that, the triage algorithm was tested by comparing the triage results of the proposed algorithm with triaging by experts.

Results: Six factors were selected as key attributes, i.e. fatal outcome, serious ADRs, positive rechallenge, new drug, change in reporting and sources of reports. Four attributes used in the Thai Signal Detection Program were excluded, i.e. the drug-ADR associations, WHO-ART critical term, disproportionality and volume of reports. Experts gave the weight for the six key attributes using their experiences. To test the proposed algorithm, systemic antibiotics with 86 SDRs in total were triaged by SDAWG and eight SDRs were further assessed. Six SDRs were consistent with the result of the proposed signal triage algorithm (75% agreement) that was the top of the priority ranking. The other two SDRs were selected by SDAWG because of the highly-concerned, serious ADR and unfamiliar case. These could be because of the drug or ADRs for the current interest, level of key attributes, comorbidity and concurrent medication use, and characteristics of experts' opinions.

Conclusion: The signal triage algorithm can enhance the efficiency of the triage method by experts, as it is systematic, transparent, timely, repeatable and also scientifically based. More research is necessary to evaluate and/or improve this triage algorithm.