Estimands in Safety Analytics

Problem Statement

The value of well-defined safety estimands is not generally appreciated and has not been communicated well to safety scientists.

Safety analyses are often ambiguous and, in some cases, carry the risk of being misinterpreted.
A better understanding and consequently a more consistent application of a robust estimand framework in the analysis of safety data would eliminate this ambiguity and drive clarity on both the specific question an analysis addresses, as well as the answer to that question.

Project Scope

Develop a user-friendly, one-stop education portal on the topic of safety estimands.
Encompass and reference the learnings and recommendations of other external groups working on safety estimands – work in a complementary and comprehensive fashion, and ensure information is presented in a digestible, fit-for-application manner.
Develop an education series with the goals of (1) substantially increasing the comfort level of safety scientists when articulating appropriate safety estimands (2) increasing their use in practice and (3) increasing the breadth of estimators routinely used in safety analytics.

Project Leads	Emails
Michael Colopy, Polaris Data Labs	mike@polarisdatalabs.com
Andreas Sashegyi, Lilly	aisasheg@lilly.com
Nicola Newton, PHUSE Project Coordinator	nicky@phuse.global

Objectives and Deliverables	Timetables
Series of Training Modules	TBC
White Paper	TBC

CURRENT STATUS (Updated Quarterly)

Continued progress on self-study materials by all four sub-teams, addressing the four key topic areas identified
Draft slide decks available for three of the sub teams; content staging nearly complete for the fourth
Poster presentation of the team's work at the 2025 EFSPI regulatory statistics workshop (September)
Self-study materials in peri-final form for all four sub-teams
Begin video and webinar preparations