((A-239)) (A-239) Risk-Based Stratification of Pharmacovigilance Data for a Differentiated Approach to Follow-Up

Background: Individual Case Safety Reports (ICSRs) have varying impact on Benefit-Risk (B-R). For cases with high potential to influence B-R, efficient follow-up is crucial for accurate assessment.

Objectives: To quantify the impact of stratification of ICSRs to support a risk-based follow-up approach and implications to B-R.

Methods: Spontaneous ICSRs between Jan 2019- Dec 2023 within the GSK Safety Database were stratified into the below risk-based groupings.
• Ability to obtain follow-up;
• Ability to impact B-R: (e.g., unlisted and serious);
• Case clinical utility: would case, even with follow-up, contain the required data for assessment. The Clinical Utility Score for Prioritisation (CUSP; Kara et al 2023) methodology with a score of 17 (mean score of low utility cases from study) was utilised;
• Event type: Does the case involve DMEs, product-specific events or subject to monitoring beyond routine pharmacovigilance (PV).

Possible methods for follow-up include:
• Automated: missing information is systematically identified for follow-up;
• Algorithmic: A PV expert refines the automated output eliminating unnecessary questions;
• Focused: bespoke requests tailored to the drug-event (e.g., targeted follow-up questionnaire); Active Engagement: Active conversation with patient/HCP via various digital channels (e.g., social media, EMR, etc.) to ensure highly focused, bespoke data can be obtained or generation of RWE to enrich follow-up.

Results: • 878,152 spontaneous cases identified;
• 30.8% (270,130 cases) could not be followed up for additional information due to various reasons;
• 11.4% (99,944 cases) are listed events where for some events, where monitoring using RWD may be more suitable; therefore an automated approach to follow-up would be appropriate;
• 46.5% (408,151 cases) unlikely to be of sufficient quality (CUSP < 17) to allow for proper assessment that could also be considered for automated approaches;
• 7.1% (62,693 cases) has the potential (CUSP>17) to be of sufficient quality to allow for proper assessment, which might benefit from an algorithmic approach;
• 3.7% (32,763 cases) involve DME (unlisted events with CUSP>17) and may benefit from a focused follow-up strategy. An ad hoc assessment of 123 cases demonstrated 1.8 attributes were needed to support accurate assessment.
• 0.5% (4,471) have explicit high value (DME, CUSP>17, extra risk monitoring) may benefit from an active engagement approach.

Conclusions: A risk-based approach to follow-up would allow for resources to be focused in areas that are most likely to impact B-R. This could include the generation and use of non-ICSR data (e.g., evidence from prospective or retrospective RWD efforts) to monitor and contextualize data that may be missing or incompletely captured in ICSRs.