Skip to main content
ICPE 2023 Main banner
Final Program


ICPE 2023 Program
Icon Legend
This session is not in your schedule.
This session is in your schedule. Click again to remove it.
Back
Video

Skills Course

Session: Thursday Afternoon Pre-conference Skills Courses

Advanced Methods II: How to Make Sure Immortal Time Never Happens

Thursday, August 24, 2023
2:00 PM – 6:00 PM ADT
Location: Argyle 1
Publication Number: 24
‎1. Course Aim ‎
Immortal time bias is the result of a historical misunderstanding. This bias arises when incorrect ‎methods are used to compare treatment strategies that are sustained over time and that are identical at ‎the start of follow-up. For example, when interested in the effect of treatment duration on survival ‎outcomes, assigning individuals who took treatment for 3 years to the strategy “3 years of treatment” ‎will introduce immortal time bias because only people who live for a long time can receive treatment ‎for a long time. The bias never occurs in causal analyses of observational data when a target trial is ‎explicitly emulated. This course describes how to do so.‎

‎2. Requisites Statement ‎
This course is intended for researchers who work with healthcare databases to study the comparative effectiveness ‎and safety of pharmacological interventions. There are no precourse materials to be reviewed. ‎

‎3. Course Objectives
‎a.‎ Learn the structure of immortal time bias using causal diagrams ‎
‎b.‎ Learn how target trial emulation ensures that immortal time is never introduced in data analysis
‎c.‎ Learn data analysis techniques to correctly compare sustained treatment strategies ‎

‎4. Syllabus Outline‎
The course has three sessions. 1) Emulation of target trials with sustained treatment strategies, 2) ‎Cloning, censoring and weighting, and 3) An application to real world data. The first two sessions ‎present the theory in accessible language and with simplified examples. The methodology introduced ‎here has three steps. The first step is cloning people to assign them to multiple treatment strategies. The ‎second step is censoring clones when they deviate from their assigned treatment strategy. The third step ‎is performing inverse probability weighting to adjust for the potential selection bias introduced by ‎censoring. The third section describes a real world practical application and provides a step-by-step ‎algorithm for implementation of the methods. ‎