During the medical review of Investigational New Drug (IND) SUSARs, FDA mandates sponsors to diligently identify analogous suspected adverse events previously documented and systematically evaluate the current safety report within the context of these previously submitted cases. This analytical process is referred to as Analysis of Similar Events (AOSE). To achieve this, sponsors typically employ queries within the safety database to retrieve pertinent data from the target cases. They then proceed to conduct a meticulous manual comparison to uncover meaningful insights.
An example of a Typical Analysis Outcome.
“Search using PT Heart Rate decreased after using Company DRUG XXXXXX was conducted on Safety Database for all the events reported until 04 April 2024. The search yielded 56 cases. The cases in question involved patients between the age group of 20 to 40 years. It includes 20 male and 36 female patients. Outcome was Not recovered for 15 and recovering in 11 cases and unknown for 30 cases. Dechallenge was negative in 10 cases, not applicable in 27, and unknown for rest and rechallenge was not applicable in 49 cases, negative in 1 and unknown in rest remaining cases, based on above data the risk of usage of the DRUG XXXXX is assessed as ……….”
As the volume of Individual Case Safety Reports (ICSRs) escalates, the need to efficiently extract significant insights becomes increasingly pronounced. Advanced analytical techniques offer an effective solution in this regard. One such technique is Affinity Analysis.
Affinity Analysis: An Overview
Affinity Analysis, also recognized as Market Basket Analysis, constitutes a data mining methodology extensively utilized in the retail sector to unveil correlations and patterns among items frequently bought in conjunction. This exploration of connections offers valuable insights that aid in making informed decisions about the arrangement of products on supermarket shelves. For instance, recognizing that customers often purchase bread alongside jam prompts the strategic placement of these items in proximity. This optimization strategy is designed to boost sales by capitalizing on consumer preferences. Despite its apparent contrast with medical reviews, the ingenious application of this technique can elevate the analysis of Individual Case Safety Reports (ICSRs) within the context of Analysis of Similar Events (AOSE).
Applying Affinity Analysis to AOSE in ICSR Medical Review
In Affinity Analysis, items are linked based on co-occurrence. In ICSRs, Adverse Events, Drugs, Patient Demographics, Medical History etc. are akin to items. Defining the data set as all the cases with similar events and by treating individual ICSR data elements as “items,” AOSE can uncover hidden patterns and correlations within ICSRs. These patterns may signify potential drug interactions, synergistic effects, or specific patient populations prone to certain adverse events.
Just as Affinity Analysis identifies items frequently purchased together, it can help in identifying correlations between various dimensions in an ICSR that frequently occur together for a given drug event pair in ICSRs. This approach can assist medical professionals in pinpointing the potential root causes of complex adverse event scenarios.
Conclusion
The marriage of Affinity Analysis and AOSE opens exciting possibilities for enhancing the medical review of ICSRs and providing a novel approach to pharmacovigilance. While this blog focuses on a specific use case of AOSE, this innovative application could be leveraged to provide real time safety insights on the screen during the review, empowering medical professionals to perform a guided and informed medical review and make data-driven decisions, ultimately contributing to safer medical products and improved patient outcomes. With the advent of cloud computing and seemingly limitless computing power it brings with it, real-time application of such algorithms is becoming a reality. We at Ultragenic are focused on this area and would entertain a conversation with you about specific application scenarios.
